Species. New Interdisciplinary Essays

Acknowledgments

Initial work on this volume was supportedby SSHRCgrant number 410-960497 awarded by the Social Sciencesand Humanities ResearchCouncil of ' Canadawhile I was at Queens University ; I completedthe volume with the assistanceof a ResearchBoard Grant Horn the University of Illinois, UrbanaChampaign. I would like to thank Allison Dawe for her researchassistance with the project in its early stages; Peter Asaro fQr his assistancein its latter stages; and Unda May for her masterly, efficient, and patient work in formatting the manuscript. Although it should almost gQ without saying, I would especially like to thank all the contributors to the volume, not simply for collectively being the sinequa non for the volume, but for making the editorial aspectsof the project less chorelike than they might have been. (SometimesI even had the illusion that I was having fun.) Apart Horn providing the essaysthemselves, ' many contributors also read and offered comments on other contributors essaysand gave generaladvice, as needed, about the constitution of the volume as a whole.

Contributors

Scott Atran holds appointments at the Centre National de la Recherche Scientifique(CNRS- CREA) in Parisand the University of Michigan, Ann Arbor . He is the author of CognitiveFoundationsof Natural History (Cambridge University Press, 1990) and the editor, with Douglas Medin, of Folkbiology (MIT Press, 1999). Richard Boyd is a professor of philosophy at Cornell University . He has authored paperson scienti6cand moral realism, and is the editor, with Philip ( MIT Press, 1991). Gasperand J. D. Trout , of ThePhilosophyof Science Kevin de Queiroz is an associatecurator and zoologist in the Department of Vertebrate Zoology at the National Museum of Natural History of the SmithsonianInstitution. He was previously a Tilton Postdoctoral Fellow at . He has published both empirical papers the California Academy of Sciences on the evolution and systematicsof vertebratesand theoreticalpaperson the methods and philosophy of systematic biology , including several previous papers on species concepts. He is currently studying the phylogeny of Holbrookiaand Anolis lizards and is developing a phylogenetic approach to biological nomenclature. John Dupre holds appointmentsat Birkbeck College, London, and the University of Exeter, having taught previously at Stanford University . In 1993, he published TheDisorderof Things: MetaphysicalFoundationsof the Disunity (Harvard University Press). of Science Marc Ereshefsky is an associateprofessor in the Department of Philosophy at the University of Calgary. He has published a number of papers on speciesand is the editor of The Units of Evolution: Essayson the Nature of ( MIT Press, 1992). Species . Paul E. Griffiths is senior lecturer in the Unit for History and Philosophy of Science , University of Sydney, Australia. He previously taught at Otago University in New Zealand. His publications include What EmotionsReally Are: The Problemof PsychologicalCategories(University of Chicago Press,

1997) and, with Kim Sterelny, Sexand Death: An Introductionto the Philosophy of Biology(University of Chicago Press, 1999). David L. Hull is the Dressier Professor in the HumaniHes in Weinberg College at Northwestern University, Evanston, Illinois. He is the author of Darwin and His Critics (Harvard University Press, 1973; reprinted by the University of Chicago Press, 1983), Philosophyof BiologicalScience(PrenticeHall, 1974), Scienceas a Process(University of Chicago Press, 1988), The Metaphysicsof Evolution (State University of New York Press, 1989), and Science and Selection(Cambridge University Press, 1998). He is editor of the seriesScience and Its ConceptualFoundations at University of Chicago Press. Frank C. Keil is a professor in the Department of Psychology at Yale University . He taught for twenty years at Cornell University, is the author of Semanticand ConceptualDevelopment : An Ontological Perspective(Harvard , Kinds, and CognitiveDevelopment University Press, 1979) and Concepts ( MIT Press, 1989), and is the coeditor, with Robert Wilson, of TheMIT Encyclopedia of the CognitiveSciences (MIT Press, 1999). Brent D . Mishler is a professor in the Department of Integrative Biology and Director of the University and Jepson Herbaria at the University of California, Berkeley. He receivedhis PhiD. from Harvard University in 1984 and was in the faculty at Duke University unHl1993 , when he moved to UC Berkeley. He is a systemaHst specializing in mossesand has thus grappled with speciesboth in practice and in theory. He has published extensively on species concepts as well as on phylogeneHcs at many scales, from moss speciesto the overall relaHonshipsof the green plants. David L. Nanney becamea professor in the School of Life Scienceat the University of Illinois in Urbana-Champaignin 1959. He becamean emeritus professor in the Department of Ecology, Ethology, and Evolution in 1991. He receivedhis graduatetraining under Tracy Sonnebornat Indiana University (PhiD., 1951) and taught at the University of Michigan before moving to Illinois. His studieshave focusedon the domesticationof the ciliated protozoan Tetrahymenathermophiiaas a laboratory instrument and have included . genetic, developmental, and evoluHonary mechanisms Daniel C. Richard son is a graduate student in the Department of Psychology at Cornell University . Kim Sterelny is an Australian now living in exile in New Zealand. StarHng as a philosopher of language and mind, he has become increasingly interested in the philosophy of biology . He retains his original interests, but now ~ th a strongly evolutionary spin. He is coauthor, with Paul Griffiths, of Sex ' and Death: An Introductionto the Philosophyof Biology, due to appearwith the . University of Chicago Pressin 1999.

Contri~ ors

Robert A . Wilson is an associateprofessor in the Department of Philosophy and a member of the Cognitive ScienceGroup at the BeckmanInstitute at the University of Illinois, Urbana-Champaign. He taught previously at ' Queens University , Canada. He is the author of CartesianPsychologyand of the Mind (Cambridge University PhysicalMinds: Individualismand the Sciences Press, 1995), and is the general editor, with Frank Keil, of TheMIT ( MIT Press, 1999). of the CognitiveSciences Encyclopedia

Introduction

This volume of twelve specially commissionedessaysabout speciesdraws on the perspectivesof prominent researchersfrom anthropology, botany, developmental psychology, the philosophy of biology and science, protozoology , and zoology. The concept of specieshas played a focal role in both evolutionary biology and the philosophy of biology , and the last decadehas seen something of a publication boom on the topic (e.g., Otte and Endier 1989; Ereshefsky1992b; Paterson1994; Lambert and Spence1995; Claridge, Dawah, and Wilson 1997; Wheeler and Meier 1999; Howard and Berlocher : New InterdisciplinaryEssaysis distinguished from other recent 1998). Species collections on speciesand the speciesproblem in two ways. First, by attempting to be more explicitly integrative and analytical, this volume looks to go beyond both the exploration of the detailed implications of any single speciesconceptd . Lambertand Spence1995, and Wheeler and Meier 1999) and the survey of the ways in which speciesare conceptualized by researchersin various parts of biology (d . Claridge, Dawah, and Wilson 1997). As a whole, it takes a step back from much of the biological nitty gritty that forms the core of these recent books on speciesin order to gain some focus on general claims about and views of species. Authors for the current volume were explicitly encouragedto addresssome subset of five general themesthat tied their particular discussionsto broader issuesabout specieswith a philosophical edge to them. Half the contributors have their primary training in philosophy. The volume is thus deliberately more philo sophicalin its orientation and in the content of the essays.Yet the biological : New InterdisciplinaryEssaysis, I believe, rich enough for the detail in Species volume as a whole to contribute both to the philosophy of biology and to evolutionary biology itself. Second, the volume addshistorical and psychologicaldimensionstypically missing from contemporary discussionsof species. The historical slant is reflectedin essaysthat considerthe Linnaeanhierarchy (e.g., Ereshefsky) and the Modem Synthesis (e.g ., Nanney), as well as in those essaysthat draw on more general consIderationsfrom the philosophy of science(e.g ., Boyd) . and in those that offer particular solutions to the speciesproblem (e.g ., de Queiroz). Although the principal purposeof theseessaysis not to contribute

to the history of biology , they are often able to appeal to that history in order to enrich our understanding of speciesand the biological world. The psychologicalperspectiveis most explicit in the essaysby Atran and by Keil and Richardson, but also underlies central argumentsin severalother papers (e.g ., by Wilson and by Griffiths). Together, thesetwo featuresof the volume provide for a broad perspectiveon speciesand on the issuesin the philosophy of biology and in biology proper to which speciesare central. The papers have been organized into five sections that seemed to me to represent the most cohesive clusters of views and the most interesting sequenceof papers to read from beginning to end. Those sections are: " Monism , Pluralism, Unity and Diversity " ; " Speciesand Life' s Complications " " " " ; Rethinking Natural Kinds ; Speciesin Mind and Culture" ; and " SpeciesBegone!" The rest of this introduction mainly provides an overview of the papers in the order that they appear. There are, of course, other thematic commonalities, shared perspectives, and oppositions that this organization (or any single artifactual classificationscheme, suchas a table of contents) will obscure. One alternative way of thematically locating particular papersin the volume and of viewing the orientation of the volume as a whole is to considerthe five themesthat authors were invited to addressand the pair of themeseachpaper concentrateson most intensely. Those themes, ranked in order '~ m those that feature in the highest number of papers to those that feature in the smallestnumber, together with someaccompanying questions, are: 1

Unity , Integration , and Pluralism

. Given the

proliferation of speciesconceptsin recent years, how should these be viewed? In what ways do they compete with one another? concepts Which proposals should be seen as the main contenders for " the" species concept, and by which criteria should they be evaluated? What are the prospects for developing an integrated speciesconcept? Should one be a pluralist about species ? [ Dupre, Hull , de Queiroz, Boyd, Wilson, Atran , Mishler] 1.

Species Realism

What sort of realism ? In , if any, should one adopt with respect to species what ways does our answer to this question both reflect and influence our view of other elementsin the Unnaean hierarchy? What interplay is there between a stanceon the realism issueand broader issuesin both the philos. ophy of biology and the philosophy of science more generally? [Dupre, Sterelny, Boyd, Wilson, Griffiths, Keil and Richardson, Ereshefsky]

3 Pradicallmport In what ways are answers to the questions asked under the other four themes important for the practice of evolutionary biology and related

/ Introduction

sciences ? Should we view the resolution of the cluster of issuesoften called lIthe species problem" as foundational in some way? To what extent is the speciesproblem (merely) definitional? What is the relationship between the speciesproblem and empirical practice within the bi.ological(and other) sciences ? [Hull , de Queiroz, NanneyI SterelnyI Griffiths, Mishler] 4

Historical

Dimensions

In what ways are the views of major historical figures or movementsin evolutionary biology of significancefor contemporary views of species1Is our own view of important historical episodes(e.g., formation of the Linnaean hierarchy, the Modem Synthesis) skewed in important ways1 How can we shedlight on contemporarydiscussionsby reflecting on the recent history of evolutionary biology1 [ Nanney, Ereshefsky]

5 Cognitive Underpinnings To what extent do the literature on children' s naive biology and anthropological work in cross-cultural psychology support nativist and universalist views of species1What fruitful interplay exists between explorations of the mental representationof biological knowledge and the philosophy of biology as it has been traditionally circumscribed1[Atran , Keil and Richardson] The summariesof the sectionsand essaysindicate that many other issues are raised in Species : New InterdisciplinaryEssays , including the plausibility of the individuality thesis about species the death of essentialism , , the interplay between ecology and evolution, the relationship betweencommon senseand scientific taxonomies, and the challenge that recent developmental systems : New theory posesto taxonomy in terms of evolutionary homologies. Species all . advances debate about of these issues Between the InterdisciplinaryEssays overviews of contemporary debates and the novel insights provided in many of the essays, the volume should prove invaluable for professionals working in the contributing fields and useful for advanced undergraduate and graduate coursesin either the foundations of evolutionary biology or the philosophy of biology . Let me turn more directly to the individual essaysand the sections into " which they are organized, beginning with Monism , Pluralism, Unity , and " Diversity , containing papers by John Dupre, David Hull , and Kevin de " Queiroz. As the title of his essay ( On the Impossibility of a Monistic " Account of Species ) suggests, Dupre argues for the rejection of monism bout . He claims , moreover, that this conclusionis the proper one to species ~ draw from the completeassimilationof Darwin' s insights about the organizati9n of the biological world . There are no perfectly sharpboundariesbetween preexisting natural kinds species that would allow for a monistic account of species. Rather, what we find when we investigate the biological world is / Introduction

diversity, and our schemesof classificationshould reflect both this diversity and our various theoretical and practical ways of exploring the biological world. As well as recounting familiar objections to the numerousattempts to , Dupre also offers novel responsesto provide a monistic account of species someputative problems facing pluralism. However, he tempershis pluralism, and the acknowledgment that our taxonomic system is the product of a highly contingent process, with a concessionto monism: becauseone of the points of biological taxonomy is to facilitate communicationbetween scientists , we ought to view speciesas the basal unit in one overarching taxonomic hierarchy. Thus dispensing with overlapping taxonomies, this view ' representsa less radical version of pluralism than Dupre himself has advocated in the past (e.g., Dupre 1993; d . Kitcher 1984). Hull is more scepticalabout the prospectsfor pluralistic accountsof species in his essay, " On the Plurality of Species: Questioning the Party Line." After sketching some broad issuesthat arise more generally with "respect to pluralist views, he turns to examine some of the prominent expressionsof pluralism by Kitcher (1984), Ereshefsky(1992a), and Stanford (1995). Hull then turns to comparehis own (1997) attempts to classify and evaluate the ' plethora of speciesconceptswith Mayden s attempts (1997). WhereasHull " " reachedthe grudging conclusion, as he calls it here, in 1997 that no one speciesconcept won out within the criteria he proposed, Mayden arrived at a form of monism. Returning to one of the earlier themesof his essay, Hull attributes the differencehere in part to the fact that as a practicing scientist, Mayden has to make more de6nitive theoretical commitments than a philosopher who stands outside the practice of scienceand surveys options. Stanceson the pluralism issue typically reflect social and institutional facts about the advocatesof those stances , rather than biashee views argued out Horn first principles. The third essayin this section, de Queiroz' s "The General Lineage Concept of Speciesand the DeMing Properties of the SpeciesCategory," develops a solution to the speciesproblem that de Queiroz has recently (1998) defended: what he calls the generallineageconcept of species. This view equatesspecieswith segmentsof population lineages, and de Queiroz argues not only that it underlies " virtually all modem ideasabout species," but that it illuminates a wide range of issuesabout species, including debatesabout speciation, the individuality thesis, and speciesrealism. de Queiroz also proposes that it allows one to dissolve the debatebetween monists and pluralists. He continues by tracing the history of the population lineage concept Horn Darwin through the early part of the Modem Synthesisin the work of . Huxley, Mayr , Dobzhansky, and Wright , to the more explicitly lineagefocused concepts of Simpson , Hennig, and Wiley . Although this historical sketchconstitutes a minor part of de Queiroz' s wide-ranging essay, it serves to buttress his proposals about the ways in which the population lineage concept underlies many apparent disagreementsbetween advocatesof different - speciesconcepts. The paper concludeswith a philosophical diagnosis

/ Introduction

of why this underlying unity hasbeenlargely unrecognizedin contemporary debates. The two papers in the next section, " Speciesand Life' s Complications," look at very different issuesthat arise for particular speciesconcepts. David ' " " ' , probes Mayr s Nanney s When Is a Rose?: The Kinds of Tetrahymena biological speciesconcept (BSC) and the notions that it employs, such as a closedgene pool, from the perspectiveof a longtime practicing ciliate biologist . Nanney conveys interesting information about the ciliates (Tetrahymena in particular)- such as the relative independenceof genetic and morphological subdivisions, and the clonal propagation of these ancient protists (some of which include asexually reproducing populations)- that pose problems for the BSC; he also revealsenough of the history of protozoology to suggest why the Aeld has a strained relationship to the Modem Synthesisand concepts forged during it . One striking conclusion of the essay is that microbiology, having essentially bypassedthe Modem Synthesis, awaits a new synthesisthat focuseson more than the most recent snapshotof a history of life that stretches back almost four billion years. Kim Sterelny's essay, " Speciesas EcologicalMosaics," offers a defenseof a form of realism about speciescommitted neither to universalismabout any speciesconcept or deAnition nor to any type of speciesselection. Some(but not all) speciesform what Sterelny calls ecological mosaics ; which are madeup of ecologically diverse populations of organisms. As structured and diverse metapopulations, such mosaics are subject to evolutionary change when there is an ecological or geographic fracturing of the metapopulation, but ' they are also stabilized by what he calls Mayr s Brake, the mechanismsof reproductive isolation central to Mayr' s well-known account of speciation. ' Sterelny explores this idea through a discussion of V rba s and Eldredges views of evolutionary change in which he argues, amongst other things, that those views should be divorced from their authors' own fondness for Paterson's (1985) recognition concept of species. Sterelny's scepticismabout universalism and thus monism draws on the claim that, like organisms, speciesand the complex, ecological organization they possesswere invented at some point in evolutionary time, forming a grade of biological organization that, like organismal individuality , only some clusters of biological entities have. The essaysby Richard Boyd, Paul Griffiths, and Rob Wilson in the next " section, Rethinking Natural Kinds," revive the ontological issueof whether speciesare natural kinds or individuals by offering a reexamination of the " notion of a natural kind. Boyd' s paper, Homeostasis, Species , and Higher " Taxa, develops the conception of homeostaticproperly cluster(HPC) kinds that he briefly introduced in earlier work (1988, 1991). Here, Boyd both provides the broader philosophical context in which that conception funcons and shows how it applies to several issuesconcerning species. More ~ speciAcally, he defends the idea that HPC kinds are an integral part of an overall, ~ealist view of sciencethat accommodatesthe inexactitude, natural

Introdu~

, and historicity of many sciences , including the biological vagueness sciences . He then arguesthat speciesand at least some higher taxa are HPC kinds, and indicateshow his view makesplausible a form of pluralistic realism . A passing theme in the essayis that in the HPC conception of natural kinds, the contrast between natural kinds and individuals is of less importance than it is in a traditional notion of natural kinds, thus deflating the significance of the individuality thesis about speciesdefended by Hull (1978) and Ghiselin (1974, 1997), and the subsequentdebateover it. " , and Kind: ResuscitatingSpecies My own contribution- Realism, Essence " Essentialism1- takes its cue from Boyd' s earlier work on HPC kinds. After outlining how both the individuality thesis about speciestaxa and pluralism about the speciescategory have been developed becauseof problems with traditional realism, I use two examplesfrom the taxonomy of neural statesto suggestthat there is more than merely conceptualspacefor a view closer to traditional realism than either of these fairly radical proposals. This middleground position is a version of the HPC view of natural kinds, and in contrast to Boyd's own development of this view, I argue that this position is incompatible with both the individuality thesis and pluralistic realism. This essaythus steps outside of the philosophy of biology to the philosophy of psychology and neuroscienceto shed some light on natural kinds more generally and on realismand pluralism about speciesin particular. ' " Griffiths s " Squaring the Circle: Natural Kinds with Historical Essences looks at the treatment of the notion of natural kinds by a variety of researchersacrossthe biological sciences , including systematists(regarding taxa and ) species process structuralists (regarding developmental biology ). Griffiths defends the idea that natural kinds can have historical essences , using this idea to addressthe claim that there are no (or few) laws of nature in the biological sciences . For Griffiths, concepts of taxa and of parts and es in can be basedon the idea of an evolutionary rather than process biology a distinctly structural or developmental homology . Griffiths seesphylogenetic inertia and its basis in the developmental structure of organisms as a " mechanismfor producing what Boyd calls the " causalhomeostasis of natural kinds. The two papersin the next section, " Speciesin Mind and Culture," present perspectiveson the issues surrounding the psychological and cultural representationsof central biological concepts, such as the speciesconcept. In " The Universal Primacy of Generic Speciesin Folkbiological Taxonomy: " Implications for Human Biological, Cultural, and Scientific Evolution , Scott Atran draws on recent cross-cultural experimentalresearchwith the Maya in Guatemala and with midwestem urban college students that probes the . strength of inductive inferencesacrossvarious levels of biological categories. Atran has found surprising similarities across these forest- dwelling and urbanizedpopulations that cry out for psychological explanation. He argues for the universality acrosscultures of what he calls genericspecies , a level of

Intr~

tion

' organization in the biological world that doesnt distinguish the Linnaean speciesand genus categories; he proposes a domain specific representation -basedhabits of the of this category and explores its relationship to essence mind and the cultural development of various speciesconceptsin Western science.Atran concludeshis paper with somethoughts about recent views of pluralism and speciesand about what these views imply about the relation betweencommon senseand science. Frank Keil and Daniel Richardson discuss the psychological representation of speciesand of biological knowledge more generally in their essay, " , Stuff, and Patterns of Causation." They argue that the substantial Species developmentalliterature on biological knowledge often presentsamisleading conception of what intuitive or folkbiology must be like in order for speciesand other biological categoriesto have the distinctive psychological featuresthat they do, suggesting severalnew lines of empirical research. By " " exploring what has been called psychological essentialism about biological kinds and its relationship to essentialismin the philosophy of biology , Keil and Richardson call for more careful empirical examination of the nature of our mental representationof the biological world and identify a number " of cognitive blases that contribute to what they call the vivid illusion of " species. They claim that although speciesdo seemto have a distinctive psychological representation, the specificform that representationtakesremains an largely open empirical question. " The concluding section- SpeciesBegonel" - contains two essaysthat, in their own ways, expresssome skepticismabout the specialreality of species that is the focus of biological and philosophical controversy regarding " " species(as in the speciesproblem ). Both authors feel that speciesare as real as higher taxa, but no more than the genuses, families, orders, and so on that ' " those speciesconstitute. Marc Ereshefskys Speciesand the LinnaeanHierarchy " offers a review of our current thinking about the speciescategory, advocating a replacementof the entire Linnaeansystem of classification. Ere shefsky questionsthe distinctive reality of the speciescategory by pointing to the problems in drawing the distinction between speciesand higher taxa and by using the critiques of monistic accounts of speciesthat motivate pluralism to suggest the heterogeneity of the species category. Because the point of the Linnaeanhierarchy and the distinctions that it draws (e.g ., between speciesand higher taxa) has been lost through the Darwinian revolution , our current taxonomic practice createsproblems that alternative systems of classificationmay avoid. Ereshefskyconcludesby examining two such systems, though he acknowledgesthat any changeshould not be made lightly" . ' In Getting Rid of Speciesf Brent Mishler explores the application of , Mishler views the Linnaean phylogenetics to speciestaxa. Like Ereshefsky .hierarchy as outdated, and like de Queiroz (1992; cE. de Queiroz, chapter3 in this volume), he thinks that phylogenetic schemesof classificationare necessary . Mishler argues that taxa at all levels, including the least inclusive,

/ Introduction

should be recognizedbecauseof evidencefor monophyly . He believes that the failure of the various speciesconceptsto uniquely define the speciesrank in the phylogenetic hierarchy reflects reality, thus highlighting the need to get rid of the speciesrank altogether. Thus, a rank-free phylogenetic taxonomy should be applied consistently to all taxa, including the least inclusive. Mishler concludesby reflecting on the implications of his proposed reform on our ecological thinking about biodiversity and conservation.

REFERENCFS -McCord on moralrt Rlism Boyd, R. (1988). How to be a moralrealist. In G. Sayre . , ed., Essays . Ithaca , N.Y.: CornellUniversityPress , anti-foundationalism for naturalkinds. Philosophi , andthe enthusiasm Boyd, R. (1991). Realism calStudies 61, 127- 148. , M., H. Dawah , andM Wilson(1997 : Chapman and ) . The"nitsof biodiomit,v. London Claridge Hall. de Queirol. K. (1992). Phylogenetic definitionsandtaxonomicphilosophy . BiologyandPhiloso . 7 295 313 . , phy de Queirol. K. (1998). The generallineageconceptof &pedes , &pedesaiteria, andthe process of speciation : A conceptualunificationand terminologicalrecommendations ." In Howardand Berlodter . : Philosophical . Cambridge of things Dupre, J. (1993). Thedisorder founationsfor thedisunityof scimce . , Mass.: HarvardUniversityPress . Philosophy , M. (19928 ). Eliminativepluralism , 59, 671- 690. Ereshefsky of Science .b). Theunitsof eoolution : Ess , M., ed. (1991 . Cambridge Ereshefsky , Ilvs on the Mtureof species Mass.: MIT Press . Ghiselin , M. (1974). A radicalsolutionto the species . Systmratic Zooiogv1.3, 536- 544. problem .b. 1991 Reprintedin Ereshefsky andtheoriginsofspecies Ghiselin , M. (1997). Metaphysics . Albany, N.Y.: SUNYPress . . Berlocher HowardD . J., and SH andspeciation . Oxford: , eds. (1998). Endless fomrs: Species OxfordUnivenity Press . HulLD. (1976). Are species 1.5, 174- 191. reallyindividualsS,vstmratic Zooiogv HulL D. (1978). A matterof individuality. Philosophy 45, 335- 360. Reprintedin of Science . b. 1991 Ereshefsky HulL D. (1997). The idealspeciesde&nition andwhy we can't get it. In Claridge , Dawah , and Wilson . Kitcher,P. (1984). Species . Philosophy 51, 308- 333. Reprintedin Ereshefsky .b. 1991 of Science Lambert , D., and H. Spence . Baltimore , eds. (1995). Spttiationandtherecognition : Johns concept .HopkinsUniversityPress . Mayden. J. (1997). A hi~ y of speciesconcepts: The denouementin the sagaof the species . problem. In Claridge, Dawah, and Wilson. Otte , D., and J. Endier, ed. (1989). Spttiationand its conseq UtnCtS. Sunderland, Mass.: Sinauer.

/ Introduction

. andSpeciation . In E. Vrba (ed.) Species Paterson . H. (1985). Therecognitionconceptof species 1992band Paterson Pretoria : TransvialMuseumMonographNo. 4. Reprintedin Ereshefsky 1994. : . Baltimore : Colleded andtherecognition roritings Paterson . H. (1994). Eoolution of species concept . JohnsHopkinsUniversityPress 62, . Philosophy of Science Stanford , K. (1995). For pJuralismand againstmonismaboutspedes 70- 91. . New : A debate andphylogenetic Wheeler theory , Q., and R. Meier, eds. (1999). Species concepts . York: ColumbiaUniversityPress

xvii

Intr~

on

1

On the Impossibility

of a Monistic Account

of Species John Dupre

of true relationshipand [ i]f we can onceandfor all lay the bogeyof the existence - genealogical relationship realizethat thereare, not one, but many kinds of relationship we shall and so on , , cytologicalrelationship , morphologicalrelationship releaseourselvesfrom the bondageof the absolutein taxonomyand gain enormously . in flexibility and adaptabilityin taxonomicpractice - J. S. L. Gilmour, "The Development of Taxonomic Theory Since 1851" , is meantthe actual, or ideal, arrangement of any seriesof objects By the classification the and are like which those separationof thosewhich are unlike; the purpose of of this arrangementbeingto facilitate the operationsof the mind in clearly . conceivingand retainingin the memory, the charactersof the objectsin question other natural or series , , Thus theremay be as many classifications of of of any bodies , as they havepropertiesor relationsto oneanother, or to other things; or, again, as thereare modesin which theymay beregardedby the mind. - T. H. Huxley, Introductionto the Classification of Animals Most of the philosophical difficulties that surround the concept of species can be traced to a failure to assimilatefully the Darwinian revolution. It is ' widely recognized that Darwin s theory of evolution rendered untenable the classicalessentialistconception of species. Perfectly sharp discontinuities between unchanging natural kinds could no longer be expected. The conception of sorting organisms into speciesas a fundamentally classi6catory exercisehas neverthelesssurvived. Indeed, the concept of a speciestraditionally has been the paradigmaticunit of classi6cation. Classi6cationis centrally concerned with imposing conceptual order on diverse phenomena. ' Darwin s theory, as the title of his most famouswork indicates, is about the origins of diversity, though, so it is no surprise that the dominant task in post-Darwinian taxonomy has been to connect classi6catorysystemsto the received, Darwinian, account of the origin of diversity . Attractive though . The patterns of diversity this task undoubtedly is, it hasproved unsuccessful that evolution has produced have turned out to be enormously diverse, and in many casesthe units of evolutionary analysis have proved quite unsuitable for the basicclassi6catoryaims of taxonomy. Or so I argue. Why do we classify organisms? A natural and ancient explanation expressedclearly by , for example, Locke (1689, bk. 3, chap. S, sec. 9) and

Mill (1862)- is that we do so to facilitate the recording and communication of information. H I tell you some animal is a fox , I immediately convey a body of information about its physiology, habits, and so on. The more you know about animals or mammalsor foxes, the more information about that particular animal I convey. If organismscamein sharply distinguished natural kinds, internally homogeneous and reliably distinguishable from the membersof any other kind, then the identification of such kinds would be the unequivocal aim of taxonomy. A classificatory system that recognized such natural kinds would be unequivocally the best suited to the organization and dissemination of biological information. But this is just what Darwin has shown us we cannot expect (seee.g ., Hull 1965). In a domain of entities characterized , in part, by continuous gradation of prOperties and and varyingly sharp frequent discontinuities, matters are much lessclear. It is this fact about the biological world that makesattractive .the idea of taxonomic pluralism- the thesis that there is no uniquely correct or natural way of classifying organismsand that a variety of classificatoryschemeswill be best suited to the various theoretical and practical purposesof biology . Many biologists and philosophersappearto think that pluralistic accounts of specieswill lead us to Babel(seee.g ., Ghiselin 1997, 117- 121). Biologists, they suppose, will be unable to communicatewith one another if they are working with different speciesconcepts. In this paper, I argue that species pluralism is neverthelessunavoidable. However, I also defend a kind of minimal monism: to serve the traditional epistemic goals of classification, it is desirable to have one general set of classificatory concepts. However, this general taxonomy will need to be pragmatic and pluralistic in its theoretical bases. For specializedbiological purposes, such as the mapping of evolutionary history, it may often be necessaryto adopt specializedclassificatorysystems . My monism is merely semantic: I suggest it would be best to reserve - which is, as I have noted, the traditional the term spedes philosophical term for classificatory concepts- for the base-level categories of this general, pragmatic, taxonomy. Such an antitheoretical concept of specieswill discourage the conspicuously unsuccessfuland controversial efforts to find a solution to the " speciesproblem," and leave it to working biologists to determinethe extent ~o which they require specializedclassificatoryschemes for their particular theoretical projects. Monists, needlessto say, disagree about which actual species concept biologists should accept. The cheapestway to buy monism might be with a radically nominalistic phenetic concept, as conceived by numerical taxonomists (Sneathand Sokal1973). If biological classificationcould be conceived . as merely an exercisein recording degreesof objective similarity, then some particular degree of similarity could be defined as appropriate to the species . category. But few people now think this can be done. Philosophically, attempts to construea notion of objective similarity founder on the fact that indefinitely many aspectsof difference and of similarity can be discovered between any two objects. Some account of what makes a property bio-

/ I. Monism , Pluralism , Unity andDiversity

logically interesting is indispensable: there can be no classificationwholly innocent of theoretical contamination. Without wishing to deny that phenetic approaches to classificationhave provided both theoretical insights and practical benefits, I restrict my attention in this essayto more theoretically laden routes to speciesmonism. My conclusions, however, leave entirely open the possibility that a version of pheneticism, modified by someaccount of what kinds of properties might be most theoretically interesting, may be appropriate for important domains of biology . The classificationof bacteria is a likely example(seee.g ., Floodgate 1962 and further discussionbelow). In the section "Troubles with Monism," I trace someof the difficulties that have emerged in attempting to provide monistic accounts of taxonomy motivated by central theses about the evolutionary origins of diversity . I thereby hope to substantiatemy claim that as more has been learned about the diversity of the evolutionary process, the hopes of grounding therein a uniform account of taxonomy in general, or even the speciescategory in particular, have receded. In the final section, I outline my more constructive proposal for responding to this situation.

WITHMONISM TROUBLFS The potential conflict between two main goals of classificationhas long been recognized. The first and most traditional goal is to facilitate the communication of information or to organize the vast quantities of detailed biological information. From this point of view, a taxonomy should be constructed so that knowing the taxon to which an organismbelongs should tell us as much as possible about the properties of that organism. This goal must, of course, be qualified by pragmatic considerations. Indefinite subdivision of classifications can provide, theoretically, ever more detailed information about the individuals classified: assignmentto a subspeciesor a geographicalrace will presumablygive more information than mere assignmentto a species.As the basal taxonomic unit, the speciesshould be defined, therefore, to classify organisms at a level at which the gains from finer classificationwould be outweighed by the costs of learning or transmitting a more complicatedset of categories. If organismsvaried continuously with no sharp discontinuities, this balancing of costs and benefits would present a largely indeterminate problem. By happy chancefor many kinds of organismsthere appear to be sharp discontinuities at a relatively fine classificatory level that are much sharperthan any discontinuitiesat any lower level. To the extent that this is the case, the selection of the appropriate level for assignmentof organisms to speciesappearsunproblematic. in recent years, this goal of organizing biological information has been emphasizedmuch less than a second, that of mapping the currents of the evolutionary process. A recent anthology of biological and philosophical essayson the nature of speciescarried the title The Units of Evolutionand the subtitle Essayson the Nature of Species(Ereshefsky1992). Though the idea

:~ Dupre

e Impossibilityof a MonisticAccountof Species

that, by deAnition, speciesshould be the units of evolution is not uncontroversial , it is widely held. What is a unit of evolution? Evolutionary change is not changein the properties of any individual organism, but changeover time in the distribution of properties within some set of organisms. (We need not worry here whether these properties are conceived as genetic or phenotypic.) A unit of evolution is the set of organismsin which changesin the distribution of properties constitute a coherentevolutionary process. Becausean evolutionary changeis one with the potential to be maintained in future organisms, it is easy to seethat the temporal dimension of a unit of evolution must be defined by relations of ancestry. As long as we are concerned with biological evolution in which properties are transmitted genetically (and ignore some complexities of gene exchange in baderia), then evolution will be constrainedwithin setsof organismsdefined temporally by parent-offspring relations. We must then consider what determinesthe synchronic extent of a unit of evolution. A natural and attradive idea is that a should include all and only those organismswith actual or potential species reprodudive links to one another. This condition would determine the set of organismsamong whose descendantsa genetic changein any memberof the set might possibly be transmitted. To the extent that the biological world is characterizedby impenetrablebarriers to genetic exchange, then there will be distinct channelsdown which evolutionary changescan flow. The sets of organismsflowing down thesechannels, then, will be the units of evolution. Here, of course, is the great appeal of the so-called biologicalspeciesconcept (BSC)- until recently the dominant conception of the nature of species. According to this view, a speciesis conceivedas a group of organismswith actual or potential reprodudive links to one another and reproductively isolated from all other organisms. Recallingfor a moment my brief discussionof classificationas mere ordering of information, one might also supposethat the sharp discontinuities that (sometimes) determine the optimal level for making base-level discriminationsshould correspondto lines of reprodudive isolation. The flow of evolutionary change down reprodudively isolated channels,after all, should be expectedto lead to ever-growing morphological distinctness. Thus, the goals of representingthe evolutionary processand of optimally ordering biological phenomenawould turn out to coincide after all. Unfortunately, however, the biological world proves much messier than this picture reveals. Certainly, there are casesin which speciescan be identified with discoverablelowest-level sharp discontinuities marked by reproductive barriers. But such casesare far from universal, and the appealing picture drawn thus far has a range of important complications to which I now turn. Asexual Species A familiar objection to the BSC is that it has nothing to say about asexual species. A fully asexual organism is reproductively isolated from everything

/ Momsm, PluralismUnity andDiversity

. The leading proponent of the except its direct ancestorsand descendants BSC, Ernst Mayr , has concluded that there are, strictly speaking , no species of asexualorganisms(Mayr 1987). But asexualspeciesstill require classification, and indeed some asexualspeciesare more sharply distinguishablefrom related speciesthan are some sexual species. Moreover, asexualorganisms evolved just as surely as did sexualspecies.Thus, whichever view we take of the fundamental goal of assigning organisms to species, the exclusion of asexualorganismsshould lead us to seethe BSC as at best one speciesconcept among two or several conceptsnecessaryfor encompassingbiological reality. A more radical attempt to save the BSC is suggestedby David Hull (1989): in asexualorganisms, the speciesare simply organism lineages- that ' is, an organism and its descendants(p. 107).1 I take it that although Hull s proposal is attractive theoretically, it will divorce the identification of species in these casesfrom any practical utility in classification. It should also be noted that even this radical move may not work to give the biological speciesconcept universal applicability. In bacteria, although reproduction is asexual, various mechanismsare known by which bacteria exchangegenetic material. The pattern of relationships between bacteria is thus netlike, or reticulated, rather than treelike} Although I supposethat one might hope to identify a new speciesas originating at eachnode in the net, such an identification would imply the existenceof countlessspecies, many lasting only a few minutes or even seconds. The impracticality of this idea suggeststhat we would be better abandoningthe idea of applying the BSC, or indeed any evolutionarily based speciesconcept, to bacteria. Many bacterial taxonomists (seeNanney, this volume) indeed seemto have this inclination..3 Gene Flow beyond Sharp Discontinuities

A second familiar difficulty with the biological species concept is that apparently well-distinguishedspeciesfrequently do, in fad , exchangegenetic material. The classic illustration is American oaks (see Van Valen 1976). Various speciesof oaks appear to have coexisted in the same areas for millions of years while exchanging significant amounts of genetic material through hybridization. Ghiselin (1987) is quite happy to concludethat these oaks form a large and highly diversified species. Two responsesshould be offered to this conclusion. First, and most obviously, the need to makesucha move illustrates the divergence between this kind of theoretically driven taxonomy and the pragmatic goal of providing a maximally informative ordering of nature. This divergence may not much bother the theoretically inpined, but it does illustrate one of the ways in which we cannot both have our cakeand eat it in the way indicated in the most optimistic explication of t\le BSC.4 Second, such examplesthrow serious doubt on the central motivation for the BSC, which is that genetic isolation is a necessarycondition for a group of organismsto form a coherent unit of evolution. The example shows that different speciesof oaks have remained coherent and distinct

/ of a MonisticAccountof Species : OntheImpossibility Dupre

vehicles of evolutionary change and continuity for long periods of time. Ghiselin' s conclusionlooks like nothing more than an epicycle serving solely to protect the BSCfrom its empirical inadequacy.

The Absenceof SharpDiscontinuities In somegroups of plants and of microorganisms, and very probably in other kinds of organisms, there is considerablevariation, but no apparent sharp discontinuities. It is even tempting to suggest that within certain plant genera there are no species. A good example would be the genus Rubus, blackberries and their relatives. BecauseRubus lacks sharp differentiation between types, but admits great variation within the genus as a whole, it seemsunlikely that there could be any consensuson its subdivision into speciessIf we assumethat this lack of sharp differentiations is due, in part, to gene flow , the option is again open to call Rubusa single and highly polymorphic species. Though less objedionable than in the casewhere there are sharply distinguished types, as with oaks, this move again separatestheorydriven taxonomy from the businessof imposing useful order on biological diversity . Lack of Gene Flow within Sharply Differentiated Species A somewhat less familiar point is that a considerableamount of research has shown that often there is surprisingly little genetic flow within welldifferentiated species(Ehrlich and Raven 1969), most obviously in the case of speciesthat consist of numbers of geographically isolated populations, but that nevertheless show little or no sign of evolutionary divergence. Even within geographically continuouspopulations, however, it appearsthat genetic interchangeis often extremely local. This kind of situation puts great weight on the idea of potential genetic flow in defending the BSC. If populations are separatedby a distancewell beyond the physical powers of an organism to traverse, should their caseneverthelessbe consideredone of , the organisms potential reproduction, on the grounds that if , per impossibile were to find one another, they would be interfertile? The alternative, paralleling Ghiselin' s line on oaks, would be to insist that such apparent species consistedof numbersof sibling species, differentiated solely by their spatial separation. Again, one is led to wonder what the point of either maneuver would be. Clearly, to the extent that speciesretain their integrity despite the absenceof genetic exchange, it must be concludedthat something other than . gene interchangeexplains the coherenceof the species. Contendersfor this role in caseslike either of the kinds just consideredinclude the influence of . a . common selective regime and phyletic or developmental inertia. I might finally note that although I do not know whether any systematicattempt has been made to estimate the extent of gene flow in the genus Rubus, in the

I. M~ sm . Pluralism Unity andDiversity

likely event that the flow is quite spatially limited, the claim that the whole complex group with its virtually worldwide distribution can be seen as reproductively connectedis tenuous to say the least. The conclusion I want to draw at this point is that the BSC will frequently lead us to distinguish speciesin ways quite far removed from traditional Linnaean classi6cationand far removed from the optimal organization of taxonomic information. Moreover, the theoretical motivation for the BSC seemsseriously de6cient. The sorts of criticisms I have been enumerating above have led, however, to a decline in the extent to which the BSC is now acce}?ted, and this decline has been accompaniedby increasing interest in a rather different approach to evolutionarily centered taxonomy that can be speciesconcept(PSC). broadly classi6edunder the heading of the phylogenetic (The definite article preceding the term should not be taken too seriously here, as there are severalversions of the generalidea.) The central idea of all versions of the PSC is that species- and, in fact, higher taxa as well should be monophyletic. That is, all the membersof a speciesor of a higher taxon should be descendedfrom a common set of ancestors. An appropriate set of ancestors is one that constitutes a new . The branch of the phylogenetic tree. Sucha group is known as a stemspecies is a taxon is whether of PSC versions between merely important distinction or to stem a of descendants to contain species only particular required . The latter position is definitive of contain all and only such descendants cladism, whereasthe former, generally describedas evolutionary taxonomy, requires some further criterion for deciding which are acceptablesubsetsof .6 Two issuesarise in explicating a more detailed account of the descendants PSC. First, what constitutes the division of a lineage into two distinct lineages ? Second, what constitutes and hencequali6esa group as a stem species a lineageand its descendantsas a species(or, indeed, as any other taxonomic rank)? The traditional answer to the mst question is that a lineage has divided when two components of it are reproductively isolated from one another, but the difficulties raisedin connectionwith the BSCsuggestthat this answer is inadequate. Examplessuch as oaks suggest that reproductive isolation is not necessaryfor the diVision of a lineage, and worries about the lack of gene flow within apparently well-de6nedspeciessuggest that it is not sufficient either. An illuminating diagnosis of the difficulty here is provided by Templeton (1989), who distinguishes geneticexchangeability, the familiar graphic ability to exchange genetic material between organisms, and demo that '" extent the to two between exists which , organisms exchangeability they share the samefundamentalniche (p. 170). The problem with asexual taxa and with a variety of taxa for which gene exchangeis limited is that Tile boundaries de6ned by demographic exchangeability are broader than those de6ned by genetic exchangeability. Conversely, for casesin which well-de6ned speciespersist despite gene exchange, the boundaries de6ned

Dupre: ofthe Impossibilityof a MonisticAccountof Species

by genetic exchangeability are broader than those defined by demographic exchangeability(p. 178). In the light of these considerations, Templeton proposes the cohesion speciesconcept(CSC). It is not entirely clear how this concept should be interpreted. In the conclusion of his paper, he writes that speciesshould be defined as " the most inclusive group of organismshaving the potential for " genetic and/ or demographic exchangeability (p. 181). If we assumethat the I' " connective and/ or should be interpreted as inclusive disjunctionl this definition would suggest that the Iisyngameon" of oaks- that is, the set of distinct but hybridizing species- should be treated as a species. But it is cl"ear from earlier discussion that such an application is not what Templeton intends. Earlier, he defines the CSC as lithe most inclusive population of individuals having the potential for phenotypic cohesion through intrinsic " . cohesionmechanisms (p 168). A central and convincing motivation for this definition is the claim that a range of such mechanismspromotes phenotypic cohesion, of which genetic exchange and genetic isolation are only two. Equally important are genetic drift (cohesion through common descent), natural selection, and various ecological, developmental, and historical constraints . The basictask, according to Templeton, is to Ilidentify those mechanisms that help maintain a group as an evolutionary lineage" (p. 169). What, then, is an evolutionary lineage? The significanceof the conflicting criteria of genetic and demographic exchangeability is that they show it to be impossibleto define that lineage in terms of any unitary theoretical criterion . Rather, lineages must first be identified as cohesive groups through which evolutionary changesflow , and only then can we ask what mechanisms promote this cohesion, and to what extent the identified groups exhibit genetic or demographic exchangeability. PresumablyI this initial identification of lineagesmust be implemented by investigation of patterns of phenotypic innovation and descentover time. With the abandonmentof any generalaccountof speciationor any unitary accountof the coherenceof the species,it appearsthat specieswill be no more than whatever groups can be clearly distinguished from related or similar groups. This approachmay seemtheoretically unsatisfying, but to the extent that it reflects the fact that there are a variety of mechanismsof speciationand a variety of mechanisms whereby the coherenceof the speciesis maintained, it would also seemto be the best conceptwe can hope for. This conclusion makespressing the secondquestion distinguished above: How do we assign taxonomic rank, especially speciesrank, to a particular lineage or set of lineages? A prima facie advantage of the BSC is that it provides a clear solution to this problem: a speciesis the smallestgroup of . individuals reproductively connected(or at least potentially connected) one to another and reproductively isolated from all other individuals. The difficulty is that this definition would leave one with speciesranging from huge and diverse syngameons to clonal strains with a handful of individuals. Apar~ from the theoretical difficulties discussedabove, any connection be-

I. MonJsm . PluralismUnity and Diversity

tween the theoretical account of a species and a practically useful classification would surely be severed. The question that must be faced, then , is whether from the PSC point of view the idea that the species is the basal taxonomic unit - where taxonomy is conceived as providing a practically useful classification - can be maintained . Abandoning the BSC will take care of species that look unsatis factorily large by allowing a variety of cohesion mechanisms apart from reproductive isolation , but it will tend to imply the presence of disturbingly small species. Frequently there are clearly distinguishable groups of organisms subspecies, varieties , geographical races below the species level . There is no reason to suppose that these groups are not monophyletic and no reason to suppose that they are not , at least for the moment , evolving independently . There is no doubt that such groups are often clearly distinguishable , and indeed for many purposes classification at this level is the most important . Stebbins ( 1987, 198 ) notes , for instance, that foresters are often more concerned with geographic races than species and indeed can be hampered in their work by the confusing attachment of the same specific name to trees with quite distinct ecological properties and requirements . A judge at a dog show is not much concerned with the criteria that identify something as Canis familiaris . Such groups may go extinct , they may merge With other subgroups in the species, or they may be destined to evolve independently into full -blown species or higher taxa. Their evolutionary significance is thus unknown and unknowable . The same, of course, could be said of groups recognized as full species, though the second alternative (merging with other groups ) may be rare.

THECASEFORPLURALISM An evolutionarily based taxonomy appearsto be faced at this point with only two possibleoptions. The first is to considerspeciesasby definition the smallest units of evolution. Leaving aside the insurmountable difficulty of , my argument so far has been that .this detecting such units in many cases will a fundamental classification that is often much too fine to option provide be useful for many of the purposesfor which taxonomieshave traditionally been used.7 Mishler and Donoghue (1982) suggestthat this proposal is also " conceptuallyconfused. They argue that there are many evolutionary, genealogical units within a given lineage . . . which may be temporally and spatially " overlapping (1982, 498). They suggest, therefore, that it is an error to supposethat there is any such thing as a unique basalevolutionary unit and that the particular evolutionary unit one needsto distinguish will dependon ' the kind of enquiry with which one is engaged. If there is no unique basal unit, then there is no privileged unit and, from an evolutionary point of view, no theoretical reason to pick out any particular group as the species. Mishler and Donoghue therefore propose the second option, to " [a]pply speciesnamesat about the samelevel as we have in the past, and decouple

Dupre: On theImpossibilityof a MonisticAccountof Species

the basaltaxonomic unit from notions of ' basic' evolutionary units" (p. 497). This processinvolves seeingspecieson a par with generaand higher taxathat is, as ultimately arbitrary levels of organization, chosenon a variety of 8 pragmatic grounds. Although Mishler and Donoghue seethe speciesas an ultimately arbitrary ranking criterion, they do maintain a version of the PSC and, hence, do not seeit as arbitrary from the point of view of grouping. In fact, they endorse the strong, cladistic concept of monophyly as a condition on a group constituting a species(or, for that matter, a taxon at any other level). Their " " " pluralism, however, entails that comparative biologists must not make inferencesfrom a speciesname without consulting the systematicliterature to see what patterns of variation the name purports to represent" (p. 500). But given this degreeof pluralism, and the rejection of the attempt to equate the basal taxonomic unit with any purportedly fundamental evolutionary unit, one may reasonably wonder why it is desirableto insist neverthelesson the requirementof monophyly . I suspectthat part of the motivation for this requirementis the idea that there must be someanswer to the question what a speciesreally is. It was once, no doubt, reasonableto supposethat evolution had produced real, discrete speciesat approximately the classificatory level of the familiar Linnaeanspecies.Perhapsthis supposition was an almost inevitable consequenceof the transition from an essentialist, creationist view of nature to an evolutionary view. Acceptanceof evolutionary theory would require that it more or less serve to explain biological phenomenaas theretofore understood. Nevertheless, a further century of development of the evolutionary perspectivehas given us a radically different picture of biological diversity . The sharpnessof differentiation between kinds and the processesby which suchdifferentiation is producedand maintainedhave proved to be highly diverse. There is no reasonto supposethat evolution has provided any objectively discoverableand uniquely privileged classificationof the biological world. Why , then, should we continue to insist that evolution should provide a necessarycondition, namely monophyly, on any adequatebiological taxon? I can think of only three possible answers. First, it might be held that a better understandingof evolution is so overwhelmingly the most important biological task that any taxonomy should be directed at improving this understanding. Second, it might be thought that an evolutionarily basedtaxonomy, despite its problems, would provide the best availabletaxonomy, or at least a perfectly adequatetaxonomy, for any biological project even far removed from evolutionary concerns. Or third - and this, I suspect, is the . most influential motivation it may be held on general methodological grounds that a central concept such as the speciesmust be provided with a unitary definition. This third motivation might be grounded either in ageneral commitment to unification as a scientific desideratumor on the fear that failure to provide a unified accountof the speciescategory will lead to massive .confusion as biologists attempt to communicate with one another. I

~

sm , PluralismUnity andDiversity

argue, however, that none of theseproposedjustifications of the demandfor monophyly stand up to much critical scrutiny. The first answercan be quickly dismissed. Even as distinguishedan evolutionist as Ernst Mayr (1961) has emphasizedthe distinction between evolutionary and functional biology , the former being concernedwith questions about ultimate causation (how did a trait come to exist?), the latter with questionsof proximate causation(how does the trait develop or function in particular individuals?). Following Kitcher (1984), I prefer to distinguish these types of questionsas historical and structural. It is clear that questions about the ontogeny of the human eye, say, or about the processes by which it provides the individual with information about the environment, have little to do with questionsabout how humanscameto have the kinds of eyesthey ' have. Of course, just noting this fact doesnt show that we need a taxonomy based specifically on structural aspectsof organisms, but it does remind us that there is more to biology than evolution. A particularly salient domain, about which I say a bit more below, is ecology. We should turn, then, to the second, and more promising, line of thought. The fact that a great varietY of kinds of investigation takes place within biology certainly does not show that one schemeof classification, basedon phylogenetic methods, might not be adequateto all thesepurposes. To some degree, it should be acknowledged that this question is purely empirical: only the progress of biological enquiry can determine whether different overlapping schemesof classificationmay be needed. This point needsto be stated carefully. There is no doubt at all that interesting structural or physiological properties crosscutany possiblephylo genetically basedclassification. An investigation into the mechanicsof flight , for instance, will have relevance to and may appeal to a group of organisms that includes most (but not all) birds, bats, and a large and miscellaneousset of insects. In general, convergent evolution and the acquisition or loss of traits within any sizeable monophyletic group make it clear that no perfect coincidence between monophyletic groupings and the extension of physiologically interesting traits can be anticipated. Whether this calls for a distinct, nonphylogenetic systemof classificationis lessclear. To pursuethe examplegiven, there is no particular reasonwhy the student of flight should attach any particular significance to the miscellaneousgroup of organismsthat fly . Ecology, on the other hand, raisesmore difficult issues. Ecology, it may be said, is the microstructure of evolution. Nevertheless, it is not obvious that or even well suited to ecological evolutionarily based taxa will be ideal ' there are . categories predator, parasite, or even investigations Certainly, flyi ~g predator that are of central importanceto ecological theory and that include phyletically very diverse organisms. There is no reasonwhy phyleti caqy diverse sets of organismsmight not be homogeneous(for example as fully substitutableprey) from the perspectiveof an ecologicalmodel. On the oth~r hand, suchconceptsmay reasonably be treated as applying to a higher level of generality than the classificationof particular organisms. At a more

/ : On theImpossibilityof a MonisticAccountof Spedes Dupre

applied level of ecology, however, some kind of taxonomic schememust be applied to the particular organisms in a particular ecosystem. Ecology will often be concernedwith the trajectory of a population without addressing competition betweendifferent subgroupswithin that population. It may, that is, abstract from distinctions within a population, perhapscorresponding to distinct lineages, which could be fundamentalin understanding the longerterm evolutionary trajectory of the population. Groups of sibling species ) and may prove ecologically equivalent (or demographically exchangeable thus provide another exampleof a kind of distinction that may be phylogenetically significant, but ecologically irrelevant. On the other hand it is possible that behavioral distinctions within a phyletic taxon, perpetuated by lineages of cultural descent, might provide essential distinctions from an ecological perspective. It is at least a theoretical possibility that a group of organisms might require radically diverse classification from phyletic and ecological perspectives. Perhapsa population of rats, consisting of several related species , divide into scavengers , insectivores, herbivores, and so on in ways that do not map neatly onto the division between evolutionary lineages. Ecology may therefore, in principle at least, require either coarser or finer classificationsthan evolution, and it may need to appealto classifications that crosscutphyletic taxa.9 This distinction leads me to the third objection to pluralism, the metatheoretic desirability of a monistic taxonomy. Here, it is relevant to distinguish two possible aspectsof pluralism. One might be a taxonomic pluralist becauseone believes that different groups of organisms require different principles of classification, or one might be a pluralist becauseone thinks that the samegroup of organismsrequire classificationin different ways for different purposes. Monistic objections to the first kind of pluralism seem to me to have no merit. Taking the extreme caseof bacterial taxonomy, there seemto be very good reasonsfor doubting the possibility of a phylogenetic taxonomy. The various mechanismsof genetic transfer that occur between bacteria suggest that their phylogenetic tree should be highly reticulated, and standardconceptsof monophyly have little application to such a situation . The significanceof bacteria as pathogens, symbionts, or vital elements of ecosystemsmake the goals of classificationquite clear in many casesregardless of theseproblemswith tracing phylogenies. Of course, it is possible that new insights into bacterial evolution might neverthelessmake a phylo genetic taxonomy feasible. But no vast theoreticalproblem would be created if bacterial taxonomy appealedto different principles from those appropriate, say, to ornithology .to In this sense, the assumptionthat there is any unitary " " answer to the speciesproblem is no more than an optimistic hope. The suggestionthat the useof different taxonomic principlesmight lead to serious confusion is absurd. It is of course possible that an ornithologist might mistakenly supposethat a bacterial speciesnamereferred to a monophyletic group of organisms, just as it is possible that a nuclear physicist might sup-

/ Monism, Pluralism , Unity andDiversity

pose that the moon was a planet. Not every possible misunderstandingcan be forestalled. The danger of confusion is a more plausible concern regarding the idea that the sameorganismsmight be subject to different principles of classifica tion for different biological purposes. In one sense, I am happy to agreethat this type of confusion should be avoided. It would be undesirablefor a particular , to be variously defined and to have speciesname, say Mus musculus extensions to the taxonomic varying according theory espousedby various authors. We should aim to agree as far as possible which organisms are house mice. In the concluding section of this paper, I explain how I think such species names should be understood. If, to recall my hypothetical exampleabout rats, it proves useful to treat scavengingrats as a basickind in some ecological model, it would be misguided to insist that scavengingrats constitute a species. Equally clearly, however, this concessionto standardized terminology does not at all require that all species names be conceived as answering to the samecriterion of what it is to be a species. The other consequenceof insisting on an unambiguousinterpretation of particular speciesnamesis that we Cannotassumea priori that the canonicaltaxonomy incorporating standard speciesnameswill be suitable for all biological purposes. The question here is, again, an empirical one that depends ultimately on how orderly biological nature turns out to be. If it should prove to be disorderly in the relevant sense, then biology would prove to be a more complicated discipline than is sometimesassumed . But once again I cannot seethat any unavoidableconfusion need be introduced. CONCLUSION: A CASE FOR TAXONOMIC CONSERVATISM Many taxonomists and almost everyone who usesthe results of taxonomic work have complained about the genuine confusion causedby changesin taxonomic nomenclature. Someof thesechangesseementirely gratuitousfor example, changesin the names of taxa grounded in the unearthing of obscureprior namingsand in appealsto sometimesesotericrules of priority . Other changesare more theoretically based adjustments of the extent of particular taxa. Many suchtheoretically motivated changeshave beenalluded to in this paper. BSC-committed theorists will urge that discoveriesof substantial gene-flow betweenotherwise apparently good speciesshould lead us to apply one speciesnameto what were formerly consideredseveralspecies. Phylogenetic taxonomists will want to amend the extensionsof any higher taxa that fail their favored tests for monophyly, and strict cladists will promote the breaking up of prior " species" into various smallerunits when their favored criteria for lineage splitting demand it .II Less theoretically commi,tted taxonomists may promote the splitting or lumping of higher taxa on the basis of general principles about the degree of diversity appropriate to a particular rank.

/ : OntheImpossibilityof a Monistic Account of Species Dupre

There is no doubt that the taxonomic system we now possessis a highly contingent product of various historical processes. Walters (1961) gives a fascinating account of how the size of angiospermfamilies and genera can very largely be explained in terms of earlier biological lore available to Linnaeus. Considering the data collected by Willis (1949) in support of the idea that the large families- families, that is, with large number of genera- were those of greater evolutionary age, Walters arguescompellingly that the data much more persuasivelysupport the hypothesis that larger families are those that have been recognized for longer. Very crudely, one might explain the point by arguing that the existence of a well-recognized type provides a focus to which subsequently discovered or distinguished types can be assimilated. Thus, plants of ancient symbolic significance, such as the rose and the lily , have provided the focus for some of the largest angiosperm families, Rosaceaeand Liliaceae. Walters makes the suggestive observation that even Linnaeus, recognizing the similarities between the Rosaceousfruit trees, apple, pear, quince, and medlar (Malus, Pyrus, Cydonia,u and Mespilus), attempted to unite them into one genus, Pyrus. This attempt was unsuccessful , however, presuinably becauseof the economic significanceof these , plants and modem practice has reverted to that of the seventeenthcentury. Walters comments: " Can we doubt that, if these Rosaceousfruit trees had been unknown in Europeuntil the time of Linnaeus, we would happily have accommodatedthem in a single genus! A general feature of Walters's argument is that our taxonomic system is massively Eurocentric. The shapeof taxonomy hasbeensubstantiallydeterminedby which groups of plants were common or economicallyimportant in Europe. The crucial question, of course, is whether this bias is a matter for concern and a reasonfor expecting wholesalerevision of our taxonomic practices. To answer this question, we must have a view as to what taxonomy is for , and we come back to the major division introduced at the beginning of this essay: should we see taxonomy as answering to some uniform theoretical project or more simply as providing a general referenceschemeto enable biologists to organize and communicatethe wealth of biological information ? The central argument of this paper is that the more we have learned about the complexity of biological diversity, the clearer it has become that anyone theoretically motivated criterion for taxonomic distinctnesswill lead to taxonomic decisionsvery far removed from the desideratafor a general referencescheme. Of course, the contingenciesof taxonomic history will no doubt have led, in many instances, to a schemethat is lessthan optimal even as a mere device for organizing biological information. On the other hand, in . the absenceof a theoretical imperative for revision, it is essentialto weigh the benefits of a more logical organization of diversity against the costs of changing the extensionsof familiar terms. My intuition is that on this criterion taxonomic revisions will seldombe justified. We might begin by recalling part of Huxley' s account of the function of classification(in the epigraph to this essay): to facilitate the operations of the

/ I. Monism, Pluralism , Unity and Diversity

mind in clearly conceiving and retaining in the memory the charactersof the objects in question. Plainly to the extent that taxonomic namesare undergoing constant modification, what anyone person " conceivesand retains in the memory" will be potentially incommunicableto others, and the possibility of reliably adding further information obtained from the work of others will be constantly jeopardized. This is not to say that taxonomic revision is never justified. If a speciesis included in a genus in which it is highly anomalous , and if that speciesis much more similar to other speciesin someother genus, then the goals of organizing information will be better served by reassigningit . It is of coursealso true that monophyletic taxa will tend to be more homogeneousthan polyphyletic taxa, and that in paraphyletic taxataxa in which some of the descendantsof the common ancestorsof aparticUlar taxon are excluded- there will be often be a case, on grounds of similarity , for including the excluded parts of the lineage. My point is just that theseconsequences rather than monophyly itself shoUldprovide the motivation for taxonomic change, and the benefitsof such changemust be weighed carefully against the potential . costs. In this weighing process, the presumption that taxon namesretain constant extension shoUldprobably be kept as strong as possible to maximize the ability of biologists to maintain reliable and communicableinformation. To take perhapsthe most familiar example, it seemsto me that there is no case at all for revising the class Reptilia (reptiles) to include Aves (birds). This move is requiredby a strict cladistic conceptof monophyly becauseit is believed that birds are descendedfrom ancestralreptiles. We cannot exclude these avian ancestorsfrom the class that includes modem reptiles because crocodiles, still classedas reptiles, are believed to have diverged from the main reptilian lineage earlier than birds did. The fact remains, however, that most zoologists, I suppose, woUld consider crocodilesmuch more like other reptiles than either is like any bird. The attempt to convince the learned or the vulgar world that birds are a kind of reptile strikes me as worse than pointless. It may be said that the only important claim is that Aves shoUldbe classifiedas a lower-level taxon included within Reptilia, and that this classification has nothing to do with our common usageof the terms reptileand bird. Although it is certainly the case that scientific taxonomic terms frequently differ considerably from apparently related vernacular terms, this differentiation is a sourceof potential confusion that shoUldnot be willfully exacerbated(seeDupre 1993, ch. 1, and forthcoming). It is also unclearwhat advantageis to be gained from insisting on sucha revision. All evolutionists, I suppose, are likely to be familiar with recent thinking on the historical relatio ~ships within the main groups of vertebrates, and if they are not, their ignoranceis not likely to be relieved by terminological legislation. Similarly, experts on smaller groups of organisms will presumably be familiar with current thinking on phylogenetic relationshipswithin those groups. To celebrate every passing consensuson these matters with a change in taxonomic nomenclatureis an inexcusableimposition of a particular professional

/ Dupre: On the Impossibiliry of a Monistic Account of Speaes

perspective on the long-suffering consumers of taxonomy outside these phylogenetic debates. In conclusion, I am inclined to dissociatemyself from the strongest reading of the taxonomic pluralism I advocated earlier (1993; see also Kitcher 1984). In view of the limited successof theoreticalarticulations of the species category, it would seemto me best to return to a definition of the species as the basalunit in the taxonomic hierarchy, where the taxonomic hierarchy is considered as no more than the currently best (and minimally revised) generalpurpose referencesystem for the cataloguing of biological diversity . This system should provide a lingua franca within which evolutionists, economists, morphologists, gardeners, wildflower enthusiasts, foresters, and so on can reliably communicatewith one another. Where special studies, such as phylogeny, require different sets of categories, it would be best to avoid using the term species(the desirability of rejecting this concept is sometimesassertedby evolutionists). Of course, such specializedusers will be free to advocate changesin taxonomic usage, but should do so only in . Although I am inclined to doubt the desirability of a extreme circumstances of pluralism overlapping taxonomies, a general taxonomy will evidently draw broadly and pluralistically on a variety of considerations. Perhapsthe most important will be history, not an unattractive idea in a sciencein which evolutionary thought is so prominent: a goal of generaltaxonomy should be to preserve the biological knowledge accumulatedin libraries and human brains as far as possible. In addition, there would be a range of the morphological, phylogenetic, and ecological considerations that have figured in various monistic attempts to define the species. The importance of these considerationsmay vary greatly from one classof organismsto another. My feeble monism is my recognition of the importance of such a general reference system. My recognition of the likelihood that different enquiries may need to provide their own specializedclassificationsand my tolerance of diverse inputs into the taxonomic processwill leave serious monists in no doubt as to which side I am on. The position I am advocating provides, incidentally, a quick and possibly amicableresolution to the speciesas individuals debate. Species , I propose, are units of classificationand thereforecertainly not individuals. Lineages, on the other hand, are very plausibly best seenas individuals. Often, it may be the casethat the membersof a species(or higher taxon) are identical to the constituents of a lineage, but of course this coincidence does not make the speciesa lineage. And it is doubtful whether all species, or certainly all higher taxa, are so commensurablewith lineages. Resistanceto or even outrage at the kind of position I am advocating may derive from the feeling that I am flying in the face of Darwin. Darwin, after all, wrote a well-known book about the origin of species,and he was writing about a real biological process, not a naming convention. Of course, the problem is that Linnaeus (or for that matter Aristotle ) also talked about

/ I. Monism,Pluralism , Unity andDiversity

species and had in mind kinds , not things . Arguably , the tension between these two usages is at the root of the great philosophical perplexity that the concept of species has generated in this century . In arguing for reversion to the earlier usage of the term species , I am at least honoring conventions of priority . What I am proposing , however , is not much like a Linnaean taxon omy either . As many have observed , Darwin forced us to give up any traditionally essentialist interpretation of taxonomic categories and even any objectively determinate taxonomy . But almost a century and a half of biological work in the Darwinian paradigm have also shown us that evolution does not reliably produce units of biological organization well - suited to serve the classificatory purposes for which the concept of species was originally introduced , so perhaps rather than a reversion to Unnaeus , it would be better to see my proposal as a quasi- Hegelian synthesis . At any rate, if I seem to have been implying that Darwin may have been responsible for introducing some confusion into biology , I am sure no one will take this as more than a peccadillo in relation to his Unquestion ably positive contributions .

ACKNOWLEDGMENTS I would like to thankRobWilsonandChrisHorvathfor valuablecomments on a draft of this essay.

NOTES 1. It is not entirely clearhow to makethis idea precise. Obviously, not every organismfounds a . A natural idea is that every organism in lineage, unlessevery organismis to constitute a species any way genetically distinct from its parent should found a new lineage. Given, however, the possibility of the samepoint mutation occurring more than once, it could turn out that a set of genetically identical organismsmight constitute two or more distinct species. The proposal also leads to the surprising conclusion that the vast majority of speciesare asexual. As Hull notes, this conclusionmay mitigate the well-known difficulty in explaining the origin of sex by showing that sexual reproduction is a much rarer phenomenonthan is often ' supposed(1989, 109). I should also mention that Hull s proposal is made.in connectionwith the thesisthat speciesare individuals, and is thus not necessarilyan explicit defenseof the BSC. 2. It appearsthat the sameis probably true for somekinds of flowering plants (seeNiklas 1997, 74 fE.). 3. This claim is perhapsless true now than it was twenty years ago. An influential evolutionary classificationof bacteria was proposed by Woese (1987); see also Pace (1997). On the other hand, Gy Uenbergand others (1997) aim explicitly to producea classificationthat is optimal from an information-theoretic perspective, a goal that there is no reasonto supposewould be met by any imaginable phylogenetic scheme. Seealso Vandammeand others (1996) for a related proposal . It is clear, at any rate, that any possiblephylogenetic classificationof bacteria, if it is to be of ~ y practical use, must de6ne taxa with great clonal diversity. Gordon (1997), for instance, Collpopulations in feral mice was an increasing reports that the genotypic diversity of Escherichia function of the age of the mouse, indicating the development of distinct clones during the lifetime of the n:' use. I assumeone would not want to think of this developmentas speciation, but

/ : On the Impossibilityof a MonisticAccountof Species Dupre

given this clonal diversity, it is difficult to seehow any useful taxonomy could avoid being arbitrary from a phylogenetic perspective. The situation is still worse in view of the partially reticulate phylogeny consequenton genetic exchangebetweenbacteria. " 4. An extreme statementof this optimistic view can be found in Ruse(1981, 237) : There are different ways of breaking organisms into groups and they coincide! The genetic speciesis the morphological speciesis the reproductively isolated speciesis the group with common " ancestors. ' S. On R. fruticosus , the common blackberry, Benthamand Hooker (1926, 139) wrote: it varies considerably. The consequencehas been an excessivemultiplication of supposedspecies. .. although scarcely any two writers will be found to agree on the charactersand limits to be " " " " assignedto them. The same species is describedby Schauer(1982, 346) as aggregate, variable " with very numerous miaospecies . More optimistically, The Oxford Book of Wildjlowm " (Nicholson, Ary , and Gregory 1960) statesthat [t ]here are severalhundred speciesand hybrids " in the Rubusgroup, and only an expert can identify all of them . 6. SeeSober(1992) for a very clear exposition of this distinction. Although the debatehere is a fundamentalone, it is not of central concernto my essay. 1. SeeDavis (1918, 334- 338) for a discussionof someof the difficulties in subspeci&c classmca tion of angiosperms. 8. For further elaboration, seeMishler and Brandon(1987) . For more generalargumentsagainst any fundamentaldistinction between speciesand higher taxa. seeEreshefsky(1991 and chapter 11 in this volume) and Mishler (chpater 12 in this volume). 9. Somemore realistic exampleshave beendiscussedby Kitcher (1984). 10. For referencesto bacterialtaxonomy and brief discussion , seenote 3. 11. De Queiroz and Gauthier (1990, 1994) claim that taxonomic changesthey advocate will promote constancy of meaning, or definition, for taxa. Mammalia, for example, should be defined as the set of descendantsof the most recent common ancestor(i.e., ancestralspecies ) of monotremesand therians. The extensionof sucha term, however, will be constantly revisablein the light of changesin opinion about the details of evolutionary history . From the point of view of the consmnerof taxonomy, at least, I suggest that constancy of extension is surely more valuablethan constancyof definition. ' 12. Subsequentto Walters paper, the quince appearsto have been reconceivedas Chatnomtles (though not unanimouslyaccording to the few sourcesI consulted on this matter). This reconception effects a conjunction with the ornamentalflowering quinces. One might speculatethat the increasingobscurity of the quinceas a fruit might have exposedit to this annexation, whim one doubts could have happenedto the apple.

REFERENCES , 7th ed., revisedby, A B. Bentham , G., andJ. D. Hooker(1926). Handbook of theBritishFlorR Rendle . Ashford , Kent: L. ReeveandCo. : A discussion on taxonomicrankandaf6nity. Notes Davis, P. H. (1978). The movingstaircase fromtheRoyRI BotRnicGIlrden36, 325- 340. : Phyloge de Queiroz , K. , andJ. Gauthier(1990). Phylogenyasa centralprinciplein taxonomy 307 322 . . System RticZooiog 39 neticdefinitionsof taxonnames , ,v de Queiroz, K. , andJ. Gauthier(1994). Towarda phylogeneticsystemof biologicalnomenclature tmdEvolution . Trends in Ecology 9, 27- 31.

/ I. Monism , Pluralism Unity

and Diversity

: Metaphysical . Cambridge of things foundations of thedisunityof science Dupre, J. (1993). Thedisorder . , Mass.: HarvardUniversityPress . Cambridge ) . Are whalesfish? In D. L Medin andS. Atran, eds., Folkbiology Dupre, J. (forthcoming . , Mass.: MIT Press . Science 165, 1228- 1232. Ehrlich , P. R, andP. H. Raven(1969). Differentiationof populations 1992.) (Reprintedin Ereshefsky 58, , M. (1991). Species , highertaxa , andthe unitsof evolution. Philosophy Ereshefsky of Science 84- 101. 'on: &saysonthenatureof species . Cambridge , M., ed. (1992). Theunitsof evoluh , Mass.: Ereshefsky MIT Press . of bacterialtaxonomy . Bacteri , G. D. (1962). Someremarkson the theoreticalaspects Floodgate 26, 277- 291. ologyReview 2, Ghiselin , M. T. (1987 , individuality,andobjectivity. BiologyandPhilosophy ). Species concepts 127- 143. andtheoriginof species . Albany, N.Y.: SUNYPress . Ghiselin , M. T. (1997). Metaphysics Gilmour,J: S. L (1951). Thedevelopment of taxonomictheorysince1851.Nature168, 400- 402. Collpopulationsin feralmice. MicroGordonD. M. (1997). Thegeneticstructureof Escherichia . 143 2039 2046 biology , , J. Schindler , andM. Verlaan(1997). Classi Gyllenberg , H. G., M. Gyllenberg , T. Koski, T. Lund ficationof Enterobacteriacea . 143 721 , by minimizationof stochastic Microbiology complexity 732. on taxonomy : Two thousandyearsof stasis . British Hull, D. L (1965). Theeffectof essentialism 15, 314- 326, and16, 1- 18. (Reprintedin Ereshefsky 1992.) Journal for thePhilosophy of Science 'on. Albany, N.Y.: SUNYPress . Hull, D. L (1989). Themetaphysics of evoluh . Philosophy Kitcher,P. (1984). Species 51, 308- 333. (Reprintedin Ereshefsky 1992.) of Science humanunderstanding Locke , J. (1689; 1975edition). An essay , editedby P. H. Nidditch. concerning Oxford: OxfordUniversityPress . 134, 1501- 1506. Mayr, E. (1961). Causeandeffectin biology. Science : Scientificprogressandphilosophical termi) . The ontologicalstatusof species Mayr, E. (1987 2, 145- 166. nology. BiologyandPhilosophy . : Parker MilL J. S. (1862). A system , 5th ed. London , SonandBourn of logic : A casefor pluralism . Systemah 'c Mishler, B. D., andM. J. Donoghue(1982). Species concepts 31, 491- 503. (Reprintedin Ereshefsky 1992.) Zoology Mishler, B. D., and R N. Brandon(1987). Individuality , pluralism , and the biologicalspecies and 2 414 . . 397 , conceptBiology Philosophy . Oxford: Oxford Nicholson , B. E., S. Ary, andM. Gregory(1960). TheO:rfordbookof wildflowers . UniversityPress . Chicago : Universityof ChicagoPress . Niklas, K. J. (1997 ' onarybiologyof plants ). Theevoluh . Science 276, 734Pace .view of microbialdiversityandthebiosphere , N. R (1997 ) . A molecular 740. Ruse : Naturalkinds, individuals , M. (1987). Biologicalspecies , or what? BritishJournalfor the 38, 225- 242. (Reprintedin Ereshefsky 1992.) PhiloSophy of Science

/ Dupre: On the Impossibilityof a MonisticAccountof Species

Schauer , T. (1982). Afieldguideto thewildflowersof BritRinRndEurope . , translated by R. Pankhurst : Collins. London Sober , E. (1992). Monophyly. In E. A. Uoyd andE. F. Kelier, eds., Keyroords in eoolution Rrybiology . Cambridge : HarvardUniversityPress . Sneath , P. H. A., andR. R. Sokal(1973). NumeritRlt RXonomy . SanFrancisco : W. H. Freeman . Stebbins : Semantics andactualsituations , G. L. (1987). Species . Biology RndPhilosophy 2, concepts 198- 203. ton, A R. (1989). The meaningof speciesand speciation : A geneticperspective . In D. Temple Otte andJ. A Endier Rndits consequences . Sunder , eds., Spetiation . (Reprinted land, Mass.: Sinauer in Ereshefsky 1992.) Vandamme , P., B. Pot, M. Gillis, P. De Vos, K. Kersters taxon, andJ. Swings(1996). Polyphasic a consensus to bacterial . Rl , Reoiero 60 407 438 . omy , taxonomyMicrobiologic approach Van Valen , L. (1976). Ecologicalspecies , multispecies , oaks. Til.1'on 25, 233- 239. (Reprintedin 1992.) Ereshefsky Walters . NewPhytologist , S. M. (1961). Theshapingof angiosperm 60, 74- 84. taxonomy Willis, J. C. (1949). BirthRndspreRd . Geneva : Conservatoire et JardinBotaniquede la of pltmts Ville. Woese evolution.Microbiologi CRI Review 51, 221- 271. , C. R. (1987 ). Bacterial

/ I. Monism, Pluralism, Unity and Diversity

2

On the Plurality of Species: Questioning the

Party Line

DavidL. Hull

In the nineteenth century, one of the hot topics of debate was the plurality of worlds. Did God create a single Earth inhabited by all and only those souls that Jesusgave the opportunity to be savedfrom eternal damnation, or did He createmillions of worlds inhabited by just as many morally responsible beings? On the first alternative, God would appearto be as profligate as the most extravagant wastrel. He createdmillions of nebulae, each containing just as many stars, eachof which might have planets circling it in stately regularity, but on only one of these planets circling a single star did He breath soul into a single species.What a waste. But if we assumethat God is the ProtestantGod of " wastenot , want not," surely He would not have let so many opportunities slip through his fingers for creating subjectsto worship him. Not only did Jesuscome down to Earth to be sacrificedfor our sins, but apparently he also repeatedthis ritual in world after world after world (for a ' history of this controversy, seeDick s 1982 Plurality of Worlds). No less a figure than William Whewell entered into this debate- on the side of the monists. For fear of damaging his hard-eamed reputation as a sober seeker after truth , Whewell anonymously published The Plurality of Worlds (1853). To those who complained that God and hence nature did nothing in vain, Whewell cited all the waste that was already so apparentin this world: We reply, that to work in vain, in the senseof producing meansof life which are not used, embryos which are never vivified , germs which are not developed ; is so far from being contrary to the usualproceedingsof nature, that it is an operation which is consistently going on, in every part of nature. Of the vegetable seedswhich are produced, what an infinitely small proportion ever grow into plants! Of animal ova, how exceedinglyfew becomeanimals, in proportion to those that do not; and that are wasted, if this be waste! (p. 249) A similar question might be asked of God about speciesin general. God creatednumerousdifferent species , but did He create a singlesori of species or many differentsoris of species7Each and every organism belongs to one speciesand one speciesonly , but are all these speciesof the samesort7 Or possibly, does anyone organism belong to many different sorts of species7

Thesequestionsare also central to the presentcontroversy between monists and pluralists with respect to species. In Whewell' s day, the plurality of worlds was a very open question, but today the party line on pluralism conflicts with respectto there being a party line on species.As Sterelny (chapter in 5 this volume) remarks, "Evolutionary theory has moved close to a consensus in seeingspeciesas historical individuals (Ghiselin 1974, Hull 1978)," but how can consensusexist with respectto the ontological statusof species if pluralism is the party line among philosophers of science, especiallyphilosophers of biology ? Everyone seemsto feel obligated to espousethe position held by all thoughtful scholars- a nuancedpluralism, as distinct from any crude, simplistic monism. One problem unfortunately characteristic of such contrasts as monism versus pluralism is that the apparent differencesbetween them tend to disappear under analysis. Numerous sensesof monismblend imperceptibly into just as many sensesof pluralism. For example, Ereshefsky(1992, 688) concludes his discussion of " eliminative pluralism" with the observation that " Some may view eliminative pluralism asjust a complicatedform of monism. If that is the case, then the argumentsof this paper have been successful ." A clear contrast exists between more simplistic notions of monism and pluralism , but no one seemsto hold any of these simplistic alternatives. When pushed, most authors retreat to some platitudinous middle ground. In this respect, the issue of pluralism mirrors the conflicts over nature nurture and genetic determinism. Does anyone think that genes are sufficient for anything ? If this view is what genetic determinism entails, then genetic determinists are most noteworthy for their nonexistence. In the first section of this chapter, I discusssome very general issueswith respect to pluralism before turning to one example - biological species. I take a look at the connection between the monism versus pluralism dispute and (a) the contrast between realism and antirealism, (b) prerequisites for communication, and (c) reflexivity . I then turn to some illustrative examples - does HIV causeAIDS, and does smoking causelung cancer? Philosophers of sciencehave produced a variety of analysesof the notion of causation. Someof the cruder, simplistic analysesallow us to reject certain claimsmade by scientists who find the HIV hypotheses a conspiracy, not to mention scientistsworking for the Tobacco Institute who argue that smoking does not causelung canceror any other diseasefor that matter. However, these scientistsare able to hide behind the smoke screensgenerated " " by more sophisticated,pluralist analysesof cause producedby equally sophisticatedphilosophers. Causalsituations are so various and complicated that nothing can be identified as causing anything, just as the biological world is so varied and complicatedthat no one analysisof " species" will do. Instead, indefinitely many speciesconceptsare neededfor indefinitely many contexts. The great danger of pluralism is " anything goes." In order to avoid this end of the slippery slope, criteria must be provided for distinguishing between legitimate and illegitimate speciesconcepts. I end this paper by

/ I. Monism , Pluralism , Unity andDiversity

evaluating these evaluative criteria for various speciesdeAnitions. My conclusion is that if we retain the traditional organizational hierarchy of genes, cells, organisms, colonies, demes, species , and so on, any and all speciesdefinitions appearinadequate. One possibleway for a monist to avoid this conclusion is to abandon, not the traditional Linnaean hierarchy (Ereshefsky , 11 in this volume but the . ), chapter organizationalhierarchy

REALISM ANDANTIREALISM One reasonwhy philosophers find the monism-pluralismdebateso interesting is its apparentconnectionto the disputeover realismandantirealism . Of the four possiblecombinationsof thesephilosophicalpositions, two seem , andpluralismcombinedwith quite natural: monismcombinedwith realism antirealism . Monistsarguethat scientistsshouldstrive to find the bestway to divideup the world andthat sucha bestway doesexist, eventhoughwe ) whetheror not we have may neverknow for sure(whateverthat means found it. Yes, scientistsarefallible. Yes, conceptualrevolutionsdo occurin science , revolutionsthat reqUireus to startnot all over again, but at leasta few stepsback. Giventheseassumptions , the goal of findinga single, maximally informativeconceptualization of natureseemsboth desirableandreasonable . If one is going to be flat-footedand simpleminded with respectto onephilosophicalposition, why not two? (Holsinger[1987] interpretsSober [1984a]asa monist-realist.) , pluralistsarguethat the differencesof opinion so characteristic Conversely of sciencewill continueindefinitelyinto the future. The natureof these differences everemerge , but rarelydoesconsensus , andwhen surelychanges " the" world canbe it does, it is likely to be short-lived. How come? Because charaderizedin indefinitelymany ways, dependingon differencesin perspectives , worldviews, paradigms , and what haveyou. Eventhoughnot all of thesewaysareequallyplausible , acceptable , or promising , no oneway is clearlypreferableto all othersonceandfor all. Onemustkeepan openmind. If one is going to be sophisticated and nuancedwith respedto one philonot two? Stanford ( [1995] portrayshimselfasbeinga sophicalposition, why antirealist . ) pluralist The other two combinationsof the two philosophicaldistinctionsare somewhatstrained . It would seema bit strangeto arguethat one and only one way existsto divide up the world, but that the groupsof naturalphenomena arenot "real." They areas real producedon this conceptualization asanythingcanget! Of course , realcanbe definedin sucha way that nothing couldpossiblybe real, just asphilosophers havedefinedknowso that no one ever knowsanythingand law so that no generalizations ever countas la.ws but this as much fun as it is to seems , game, play, nonproductivein the . A combinationof pluralismandrealismseemsequallypeculiar(but extreme seeDupre1981andKitcher1984a ). Theworld canbe dividedup into kinds in numerousdifferentways, and the resultsareall equallyreal! Onceagain,

/ Hull: On the Plurality of Species

real can be deAned in such a way that two contradictory classifications of the same phenomena can refer to real groups of entities . Or one might get

incommensurable , but the sort of holistic semanticsthat generatesincommensurability again seemsa high price to pay. (For a discussionof realism and pluralism with respect to the units of selection controversy, see Sober 1984b; Sterelny and Kitcher 1988; Kitcher, Sterelny, and Waters 1990; Waters 1991; Soberand Wilson 1994; Shanahan1997.)

A RETURNTO BABEL Kitcher (1984a, 326) suggeststhat one worry possibly bothering opponents of pluralism is that it might " engendera return to Babel." Do monists treat languagemonistically1Do they think that every utterancehas one and only one meaning1Are pluralists pluralistic when it comesto language1Should I treat every utterance made by pluralists pluralistically, or is it just possible that texts do on occasionconstrain interpretations1As far as I know, no one deniesthe existenceof ambiguity and vaguenessin language. In fact, in the face of imperfect knowledge, vaguenessmay be necessaryfor communication (Rosenberg 1975). Again, I find it difficult to tell, but even the most rabid deconstructivist shies away from anything- literally anything- goes. Even they seemto assumethat they are saying something or trying to say . Possibly, they are not saying something with varying degrees of success one and only one thing with absolute clarity, but not all interpretations are equally acceptable- Emily Dickinson as a Marxist feminist. Perhapsour intended meaning is neither patent nor all there is to the story, but at the very least I have been taught that we should all aim to present our views as clearly and unambiguously as possible, and I see no reason to give up the ghost at this late date. PerhapsI can be a selective pluralist about a halfdozen conceptsat a time, but I cannot treat all of languagepluralistically all at once, not if I want to say something, not if I want other human beings to understandme.

REFLEXIVITY Early on in the prehistory of what has come to be known as the Science Wars, young sociologists who were in the midst of rediscovering epistemological relativism also stumbled upon reflexivity when an occasionalopponent askedwhy they were gathering so much evidenceto cast doubt on the . efficacy of evidence. These sociologists tried a variety of ways to extricate " " themselvesfrom this tension in their position. Some argued that sociolo" gists, when they are acting as sociologists, must treat the social world as real, and as something about which we can have sound data, whereas we should treat the natural world as something problematic- a social construct rather than as something real" (Collins 1981c, 217; see also Collins 1981a

/ I. Monism , Pluralism , Unity andDiversity

and 1981b). However, Collins also acknowledges that even sociologists periodically need to step back from their own work and treat it as problematic as well. The knowledge claims made by sociologists are also socially constructed. Becausesociologists are themselvesscientists, reflexivity poses especially poignant problems for them, but reflexivity can also be brought to bear on contrasts between object-level and metalevel positions. Philosophers are not scientists. As philosophers, we do not do science. We comment on it . (Of course, someonewho is officially a philosopher can join in the scientific ' enterprise, and vice versa.) Thus, a certain tension exists in Hempels position that data play crucial roles in science, but no role whatsoever in his logical empiricist analysisof science. Just becausescientistsclaim to explain ' phenomenain ways different from Hempel s covering-law model does not meanhis covering-law model of scientific explanation is false. If no scientists ever explained anything by deducing it from laws of nature and statements ' of particular circumstances , Hempel s covering-law analysis of explanation still could count as a totally adequateanalysisof scientificexplanation! ' Hempels holding these two positions on the role of data with respect to the object level and metalevelrespectively may seemto be a bit implausible, but it is not in the least contradictory. Parallel observationshold forplural ists. With respectto science ----so pluralists claim- serious, respectablealternative exist for every issue, but when one steps back to positions always view philosophy of science, one and only one position is acceptable:pluralism . John Maynard Smith (1998) raisesprecisely this objection in his review of Soberand Wilson (1998). " A secondreasonwhy this book is confusing is that, although the authors argue for pluralism, they are not themselvespluralists: for them, the only right way to describea model is in group selection " language (p. 640). As in the caseof Hempel's logical empiricist analysisof science, pluralists are not contradicting themselves in holding such different positions on scienceand philosophy, but it is a bit difficult to swallow their position. If one can be a monist with respect to - of all things- philosophical debates, one can certainly on occasionthink that monism is justified in science. If one and only one position is warranted with respect to the monism-pluralism dispute, then certainly one and only one position is warranted with respect to the evolutionism-creationismdispute. Possibly one and only one position is warranted with respectto the speciesproblem. If not , I am missingsomething of massivesignificance.

CONsm ERAnO NS PROFFSSIONAL . I.n general, people find .it much more plausible and desirableto counselpluralism with respect to otherpeople' s areasof expertise than their own. As I have argued above, philosophers find pluralism extremely attractive in science, -much more so than in philosophy. Scientists in turn do not find

/ Hull: On thePluralityof Species

pluralism all that attractive in their own area of expertise and usually stay well clear of philosophy. At least sometimes, scientiststhink that they have the right answer to a particular question. What would sciencebe like in the absenceof such convictions? Picture hundredsof scientists, eachbeing terribly considerateof eachother' s hypotheses: "You think that selection occurs only at the level of the genetic material? That may be so, but in addition, individual organismsare the main target of selection." " I agree with everything you say and want only to add that selection wandersup and down the organizationalhierarchy in biology ." " To supplementthe nuancedposition being expressed , I submit that no such as It selection exists. is two es not one." , really thing process " So true and selectionis of " , only peripheralimportancein evolution. " " Speakon, oh wise one. A preference for monism and pluralism waxes and wanes as various groups gain and lose power. Right now, advocatesof developmental systems theory are trying to supplant the current gene-centered world view (Moss 1992, Griffiths and Gray 1994). None too surprisingly, these developmentalists are urging pluralism. In general, groups who hold minority opinions at a particular time find pluralism to be the correct philosophical view, whereasthe groups in power are not nearly so attracted to it . During the heyday of the biological speciesconcept, Mayr saw no reason to give ground to his opponents. He insisted that there are basic units in the evolutionary process, that theseunits are delineatedin terms of reproductive isolation , and that making these units coincide with the basic units of classification is both possible and desirible. I predict that if and when developmentalistssee their views prevailing, they will ceasetheir pleas for pluralism and becomestaunchmonists. Pluralismlooks good to outsiders regardlessof whether they belong to different disciplines(e.g., philosophers looking at science) or to groups currently a minority within a particular discipline (e.g., developmentaliststrying to muscle evolutionary biologists out of their positions of power).

ANALYSFSOF CAUSA nON

Much of the problem that I am having with the monism-pluralism issue stems &om the general characterof philosophical analysesas they function in of science. At times, they seemto help; at other times, they - philosophy seemto obfuscateunderstanding. The notion of causecan serve as anillus trative example. In the early days of AIDS, scientists floundered around " trying to find lIthe causeof AIDS. Possibly drug addicts and sexually promiscuous homosexual men were battering their immune systems with so many different diseasesthat eventually their immune systemsgave up and

/ I. Monism , Pluralism , Unity andDiversity

closed down. Or perhapsan inhalant termed popperswas the cause. Or possibly some infectious agent was responsible. Rather rapidly, the evidence indicated that the most likely explanationwas also the correct explanationa virus transmitted primarily by infected needlesand sexual intercourse. At the time that scientiststurned their attention to the AIDS virus in 1982 and 1983, the evidencewas far from -conclusive. In a very real senseof the term, " " they might have beenmistaken. However, little by little a consensusemerged that AIDS is a contagious disease , and the causative agent is a virus eventually termed the human virus or HIV for short (Blattner, Gallo, and Temin 1988). immunodeficiency One virologist , Peter Duesberg (1987), disagreed. He argued that AIDS is " " not one disease , but at least four. What is mistakenly called the AIDS virus has nothing to do with the diseasesweeping through Europe and North America. Instead, the useof drugs, suchas poppers, are the real cause. Hence, the antiviral drugs suchas AZT being used to combat AIDS are doing nothing but exacerbatingthe problem. One important feature of scienceis persistence in the face of considerableopposition. On occasion, scientistswho were thought to be in the lunatic fringe in their own day eventually are vindicated . Doubting the role of HIV in causing AIDS in 1987 is one thing . Continuing to doubt it today is quite another matter (Duesberg 1996). Right now, the only explanation of AIDS in the least plausible is that it is causedby HIV . However, Duesberg (1996) usesall sorts of speciousarguments about causationto salvagehis position. Yes, yes, HIV alonecan't cause AIDS. It needshosts to invade. Without people, the various strains of virus that can infed only human beings would be in real trouble. In addition, if no one had ever isolated the blood- clotting factor in people, hemophiliacs would not have been put at greater risk than other people. If no one ever sharedneedles, if everyone used virus-impermeablerubbers whenever they , we would never have had an AIDS epidemic engaged in sex, etcetera, etcetera .I And if God were truly good, etceter, etcetera. In addition, contagious diseasesare commonly defined in terms of the presenceof the causative agent. In order to have tuberculosis, you must be infected with the tuberculosis - by definition. But it is not sufficient . In order bacterium. It is necessary for people infected by the tuberculosisbacterium to come down with tuberculosis , all sorts of other factors are also required. In sum, any philosopher worth his or her salt could explain to Duesberg that just becausethe presence ' of HIV is not sufficientfor a person coming down with AIDS, he cant concludethat it is irrelevant. The issueI wish to raiseis what advice advocatesof pluralism would have given scientistsand funding agenciesin the early 1980s, the later 1980s, and toaay. In the early 1980s, scientistsentertained all sorts of hypotheses, but by the late 1980s they had settled on HIV as the contagious agent that AIDS. If the prophets of pluralism had had the power in the late .Causes 1980s, would they have required that research time and money be distributed .equally acrossall possible causesof AIDS? Should Duesberg have

Hull:OnthePlurality ofSpecies

been funded? If so, would these pluralists be willing to live with the disastrous results of their decision? If not , why not ? How about today ? Are poppers still a plausible causative agent for AIDS ? If the contrast between monism and pluralism is to be of any significance at all , advocates on both sides of this divide have to admit that at least on occasion the position they prefer might be wrong . Monists and pluralists alike have to present a list of criteria to help in deciding when one, two , three, or more possible alternatives are justified . The danger is that every situation , no matter how apparently straightforward , turns out to be hope lessly complex . It also must be noted that the decisions that we make in this connection are likely to have effects on society . People do not , as a rule , pay excessive attention to what we philosophers have to say, but sometimes the distinctions that we make and the positions that we espouse find their way into the public at large . For example , the tobacco industry for years has claimed that the scientific data are inadequate to prove that smoking causes lung cancer. After all , some people smoke three packs of cigarettes a day and die at age ninety -five in a car accident , whereas others contract lung cancer at an early age although they have never been exposed to cigarette smoke. In most scientific contexts , the factors that scientists pick as causes are rarely necessary conditions ; they are even more rarely sufficient conditions ; and they are hardly ever both necessary and sufficient conditions . Perhaps finding necessary and sufficient conditions for the occurrence of natural phenomena is the ideal, but in most contexts we have to settle for much less. The issue is statistical correlations . Holding everything else constant , how ' much does smoking increase one s chances of contracting lung cancer- or heart disease, for that matter ? " If Smoking causes heart disease" turns out not to be true for humankind , I take it that its truth has been established for the various populations of Western countries in which it has been systematically investigated . It is now known , for example , that smoking causes heart disease for the human population of the United States. There may be inductive hazards in the extrapolation of this result to other identifiable human populations ; but it seems unnecessarily cautious to restrict the claim to an as-yet -unidentified subpopulation of inhabitants of the United States. ( Dupre 1993, 200 ) If we philosophers were to stop here, the effects of our analysis would be, from my perspective , decidedly beneficial . Smoking is a major cause of both lung cancer and heart disease, and anyone with a shred of intellectual integrity has to concur , pluralism be damned. However , we do not st~ p here. Promiscuous pluralists (not to be confused ' with Dupre s promiscuous realists ) feel obligated to resist the conclusion that . smoking causes lung cancer, even in a statistical sense, because it is insufficiently nuanced. Eve.n when people do smoke and do contract lung cancer, it " " does not follow that smoking is the cause of lung cancer. All sorts of alternatives and combinations of alternatives might and probably do playa role in people coming down with lung cancer. In fact , as the father of statis -

I. Mot : sm, Pluralism , Unity and Diversity

tics (not to mention the synthetic theory of evolution), Ronald Fisher (1958, 108) mused, perhapslung canceris one of the causesof smoking cigarettesf Causalsituations are extremely complicated. All sorts of supplementalfactors can playa role, and the reasonsfor selecting one factor and terming it " the" causewhile " demoting all the others to the position of supplemental " factors are far &om obvious. This issue arises time and again in science. Natural selectionis the causeof organic complexity, while every other influence " is a constraint." Genesare the causeof various traits, whereaseverything else is demoted to being part of the "background knowledge." Many evolutionary biologists claim that gene exchangeis crucial in the individuation " " of speciesas units of evolution, but all other factors only contribute to the cohesion that is so important in speciesbeing species. On what grounds are causesassigned to these various categories? This literature throws up a picture of the empirical world so murky that even the tobacco industry can hide in the fog.

ANDMUDDLED METAPHYSICS SPEaFS At long last, after all the precedingpreparatory discussion, it is time to turn to the speciesproblem. Numerous philosophers have urged pluralism with respectto biological species. As a point of departure, I take Philip Kitcher' s (1984a, 1984b, 1987, 1989) discussion of species. As is usually the case, ' deciding what an author s position is can pose seriousproblems. Too many alternatives present themselves. (In this respect, pluralists are correct more often than I would prefer.) With respect to his own general philosophical position, Kitcher (1984a, 308) is not very pluralistic. In his paperson species, he sets himself the dual tasks of explaining a position about speciesthat he " " terms pluralistic realism and of indicating in a general way why he thinks " that this position is true." Kitcher does not say that from someperspectives and in certain circumstance , pluralism is preferable, or that from other perspectives and in other circumstancesmonism is the correct position to hold. If I read Kitcher correctly, he believes that no form of monism is acceptable by anyone no matter what. With respect to his own philosophicaloutlook, Kitcher is inclined to monism. So am I with respectto mine. Kitcher then turns to the issue of the ontological character of species. Periodically, authors have tentatively suggestedthat speciesare like individual organismsmore than they are like universalssuchas triangularity, but until Michael Ghiselin (1974) did this position becomewidely discussed. not According to Ghiselin and later Hull (1976), if speciesare to fulfill their role in the evolutionary process, they must be conceivedof as spatiotemporally 10'caIized entities connectedin spaceand time (seealso Mayr 1999; Mishler and Theriot 1999: and.Wiley and Mayden 1999). At anyone time, species must exhibit a certain degree of cohesiveness(though the mechanismsproducing this cohesivenessmight vary ), and through time, they must be connected can be .2_They are chunks of the genealogicalnexus. The term species

Hull: On the Plurality of Species

and has been used in a variety of other senses , but when speciesare supposed to be the things that evolve, they fit more naturally in the category individual (or historical entity ) than the category class(or kind) . Kitcher rejects this position in no uncertain terms. However, in the end, our differencesseem to be primarily terminological. If only Ghiselin and I had explicated our view in terms of sets, much of the controversy would have beenavoided. Insteadof talking about individuals and spatiotemporally " unrestricted classes , we should have distinguished historically connected " sets from those sets that are not historically connected(Kitcher 1984a, 314; 1987, 186; and 1989). The notion of a classis a "bastard notion that deserves no place in anybody' s ontology . The respectableconcept is that of a set." Although I find this position too monistic, I am willing to adjust my terminology if it improves communication. However, I think that there is more to this dispute than just terminology. I am not at all sure what a historically connected sets could be. They certainly could not function in traditional extensionalset theory (Sober 1984a). Early on in this dispute, I wondered how far Kitcher was willing to extend his notion of historically .connectedsets. If he could interpret speciesas historically connectedsets of organisms, I saw no reasonwhy organismscould not be interpreted as historically connectedsetsof cells. If so, Kitcher was on the verge of everything being a set. Kitcher responded that, no, anything that might pass for internal structure in speciesis much weaker and rarer than in organisms. If you destroy the internal organization of most organisms , they die, whereasonly a relatively few specieswould go extinct if their population structure was drastically destroyed (Kitcher 1989, 186). Two points are at issue- internal organization and historical connection. Most organismsdo exhibit more internal organization than most species , but this differencesis one of degree, not kind. Most speciesdo not exhibit the internal organization common in vertebrate organisms, but the samecan be said for plants as organisms. Most plants do not exhibit the internal organization common in vertebrate organisms. They tend to be much more modular (Sterelny, chapter 5 in this volume). Holsinger (1984, 293) argues that " Although it is possible to regard a speciesas a set with a special internal structure, it is preferableto regard a speciesas an individual precisely to emphasize this internal structure" (see also Sober 1984a). What, however, are the characterand extent of this structure? Even if most speciesdo not exhibit the degree of organization that most organismsexhibit, the issueof historical connectednessremains, and it alone raisesseriousproblemsfor traditional setsimplies that conceptionsof species. Calling specieshistoricallyconnected " sets" can function in set these but to , theory my knowledge no suchversion . of set theory exists. Although Boyd (chapter6 in this volume) does not treat speciesas sets, he reasonsalong much the same lines as Kitcher. Ghiselin and I assumedthe traditional distinction between classesand individuals (see de Sousaforthcoming ). Classesare spatiotemporally unrestricted, whereas individuals are

I. Mo. {sm PluralismUnity and Diversity

spatiotemporally localized and connected. Given this fairly traditional distinction . , we argued that species are more like individuals than classes Kitcher distinguishes between two sorts of sets: those that are spatiotempor ally connectedand organized, and those that are not. Given this distinction, just about everything becomesa set, including such paradigm individuals as organisms. For Boyd (this volume), at least three different sorts of kinds exist- traditional kinds that can be defined in terms of necessaryand sufficient conditions (very rare!), kinds defined as simple cluster concepts, and homeostatic property cluster kinds. In his view, speciesare not individuals but homeostaticproperty cluster kinds. So are organisms. So is the Rock of Gibraltar! In both cases , the notions of set and kind have been so redefined that the traditional distinction betweenindividuals and setsor kinds becomes all but obliterated. Speciesare not individuals, but then neither are organisms. The key substantiveissuewith respectto Boyd' s suggestionis, of course, the nature of the mechanismsresponsible for these homeostatic property clusters, but this issue to one side, I fail to seethe advantageof the preceding maneuvers. Although the distinction between spatiotemporally unrestricted kinds and spatiotemporally localized and restricted individuals may well stem from a " profoundly outdated positivist conception of kinds" (Boyd, chapter 6 in this volume), I fail to seehow we can avoid such a distinction and still talk sense. Of course, if nothing of any importancerides on the apparent difference between " f = ma" and "Richard Nixon died before " going bald, then I must revise my understanding of science from the ground up. (For an example of the impact that a biological perspectivecan bring to the philosophical understanding of individuality , see de Sousa forthcoming, and Wilson 1999.) However, as Kitcher (1984b) makesclear, logical issuesabout set theory and mereology are not his primary concern. Instead, he objects to " genea" " " logical imperialism as the monism of the moment. For my part, if I were more of a pluralist, I would object to Kitcher' s set-theoretic imperialismas his monism of the moment and Boyd' s homeostaticproperty cluster kinds as his. In any case, Ghiselin and I argued that from the perspective of the evolutionary process, speciesmust be viewed in a particular way. Kitcher objects to that way, but even more strongly he objects to our monistic view of the evolutionary perspective. He argues at some length that the evolutionary perspectiveis not the only legitimate perspectivein biology . Ghiselin and I certainly agree. Not everyone thinks that the evolutionary perspective is " " so basic. Certainly, present-day idealists think that a scienceof form can be developed without any referenceto evolution (Webster and Goodwin 1996), and advocates of developmental systems theory would like to deth' rone evolution (Moss 1992, Griffiths and Gray 1994). However, although acknowledging other legitimate perspectivesin biology , Ghiselin and I neverthelessinsist that the evolutionary perspective is the most basic perspectivein biology (seeEldredge 1985, 200, and Ghiselin 1989, 74). Perhapsthe claim so often repeatedby evolutionary biologists

/ Hull: On thePluralityof Species

that nothing in biology makes sense except in the light of evolution is a bit of an exaggeration , but not much. In sum, I think that (a) the evolutionary " " perspective in biology is in a significant sense basic to all of biology ; (b) from the evolutionary perspective , species must be treated as historical entities ' ; and (c) Kitcher s interpretation of species as historically connected sets ' and Boyd s interpretation of species as homeostatic property cluster kinds are, to say the least, strained . Even if we accept the alternatives that Kitcher and Boyd suggest , the distinction between universals and particulars must be reintroduced by distinguishing between spatiotemporally restricted and connected sets or kinds and those sets or kinds that lack these restrictions . What such an exercise accomplish es, I fail to see.

A PLElli ORA OFSPECIES CONCEPTS All of the precedingdiscussionhas concernedvery generalphilosophical and ? Kitcher (1984a) dismetaphysical issues. How about species themselves between two families of tinguishes speciesconcepts species defined in terms of structural similarities and speciesdefined by their phylogenetic relationships(seealso Dupre, chapter 1 in this volume). He divides structural concepts into common genetic structure, common chromosomal structure, and common developmentalprogram. He divides historical conceptsaccording to whether continuity or division is paramount. He then subdivideseach of these conceptsaccording to the samethree principles of division: reproductive isolation, ecological distinctness, and morphological distinctness. The end result is nine different speciesconcepts, each one of which Kitcher thinks producesreal groups. Whether or not one is willing to go along with Kitcher' s entire classifica tion of speciesconcepts, his first division cannot be ignored. No two ways of classifying the world could be more different. On the structural similarity alternative, genesisis irrelevant. If two structures are similar enough, they are the samestructure. At some level, the vertebrate eye and the eye of the octopus are the same trait. On the historical alternative, genealogy can override similarity. The Eustachiantube in humansis the samecharacteras the spiracle in sharks. They are evolutionary homologies. The contrast is betweenhomoplasiesandhomologiesor, if you prefer, setsthat do not have to be spatiotemporally connected in any way ( homoplasies ) and sets that must be (homologies). The samedistinction also appliesto periods in human " " history- for example, between feudal and Feudal. As a homoplasy, feudal can be defined in sucha way that it can occur at different times and placesas Marxist theoriesrequire, or it canbe defined so that it is restricted to a single ' period in the developmentof Europe. If anyone finds this distinction too difficult to understand,.just compare the prices of Tiffany lamps and Tiffany style lamps. Both homoplasiesandhomologies pose serious problems. Since at least Goodman (1972), philosophershave realizedthat the notion of similarity so

I. Mo m, Pluralism , Unity andDivenity

if not pervasive in our conceptions of the world is currently unanalyzed, be askedof speciesdefinitions in terms unanalyzable. A question that must " " " is similar similar How is of similarity , enough and in what senseof similar " ? Can one level of similarity be specified- one level that can be applied classifiequally acrossall organismsto produce even a minimally acceptable cation? The answer to this question is, thus far, no. One implication of this conclusion is that no general purpose classificationof plants and animals is overallsimilarity (but possible. Generalpurposeremainsas unanalyzed, as does seeDupre, chapter 1 in this volume). The historical alternative avoids the problems that plague speciesdefinitions in terms of similarity, but it hasproblemsof its own- chiefly difficulties involved in reconstructing phylogeny, which in turn requires that homologies be distinguished from homoplasies. Organisms on Earth evolved the includes lots way that they did and no other way . Even though" phylogeny " But reconstructing of merger, there is one and only one phylogenetic tree. the information of Most . not if difficult be , can impossible extremely phylogeny extant of is world organisms. Our that we have of the biological . The principles sketchier knowledge of extinct forms .is even sketchier much of cladistic analysis were devised to establish transformation series in which genuine characters(homologies) nest perfectly and hence produce not nest perfectly is perfectly monophyletic taxa. Any characterthat does volume) argues, we this in 5 not a genuine character. As Sterelny (chapter " must view phenomenologicalspecies identifiable clustersof organisms " as fallible cluesto the existenceof evolutionarily linked metapopulations. With respect to monism versus pluralism, systematistsare put in a bind. The rules of nomenclaturedo not allow them to be pluralists. They have to as Kitcher would have produce one and only one classification not nine them do. Systematic principles that take history as basic seem appealing becausethey can promise a single classification. It may not be equally useful for all sorts of purposes, but it is at least attainablein principle. Possessinga faults, is better than single classificationas a referencesystem, no matter its . The metric having dozensof alternative and incommensurableclassifications . system of measurementis not the only possible system of measurement all for not is . It on it from purposes. God did not deliver equally good high different constructed systems very Given different contingencies, we might have . In fact, we did. We have two systemsof weight, distance, and so on ' the metric and the English. I cant speakfor others, but I find the presenceof thesetwo systemsa persistentirritation . I would hate to think how inconvenient having a dozen such systemswould be. The same goes for naming and organizing the elements in the periodic table. All sorts of different ways of organizing the elementswere proposed. The end result that any.one who ever took an introductory courseinchemis - fashion try stared at for hours on end arosein a hit or miss, highly contingent of names the . In fact, we are only now getting around to formalizing the elements with atomic numbers from 101 to 109 (see Pureand Applied

Hull: On .thePlurality of Species

Chemistry69, 2471- 2473, 1997). But having a single faulty systemthat dependsin large measureon the historicalcontingencies of its genesisis betterthanhavingno generalreferencesystemat all. Kitcher(1984a , 326327) disagrees . He thinks that we do not needto worry abouta return to Babel . Biology canand doesfunctionjust fine without sucha generalreference . Generationsof naturalhistoriansand systematists system havebeen misguidedin trying to producea single, coherent , and consistentclassifica tion of plantsand animals . Kitcherwould no doubt disagreewith a recent editorialin Nature(1997, vol. 389, 1) that took molecular biologiststo task " of for the "promiscuity the numerousconflictingnamesthat they coin for . proteinsandothermolecularstructures PRUNINGTHE TREEOF KNOWLEDGE : THE NEEDFOR CRITERIA For Kitcher(1984a ), systemsof classification are theory dependent : that is, one given theory, you might divideup your subjectmatteroneway; givena differenttheory, you might divide it up differently(see Dupre, chapter1 in this volume). I agree , but in his early work on the speciesconcept , Kitcher madeonly the faintestgesturestoward specifyingtheseallimportanttheoretical contexts . He refers to nothing more specificthan the ' legitimate interestsof biology," "biologically interestingrelations ," and so on. Sober (1984a , 334) raisesjust this complaintagainstKitcher's pluralism . Soberdistinguishesbetweenspeciesconceptsthat are currentlyplaying a significant role in active researchprogramsand species conceptsthat dependon a "mere " " hope and that lack any seriousdegreeof theoreticalarticulation ." Too manyof the speciesconceptsthat contributeto Kitcher's pluralismare of this secondsort. Theremay well be laws of form, and some very good biologistsare currentlyworking on the theoreticalbackgroundof sucha view, but it still remainslargelypromissory . Kitcheris not a total pluralist. He rejectsboth the creationistandthe phe neticspeciesconcepts , but he doesnot explicitly set out the criteriathat he hasusedin makingthesedecisions . All he saysis that we are obligatedto attention to those pay conceptsthat arebiologicallyinterestingand ignore the "suggestions of the inexpert, the inaneand the insane !" (Kitcher 1987, 190). Thosepeoplewho arepushingthe creationistspecies conceptarecertainly . I am not surehow to inexpertand inane, but they arehardly insane decidewhetheror not the creationistspeciesconceptis biologicallyinteresting . I don't think it is. Pheneticists in their turn areneitherinexpertnor insane . I also do not find their work on the speciesproblemall that inane either. Phenetic(or numerical ) taxonomistsaregenuinescientistsworkingon scientific issues in genuinely genuinelyscientificways. They also obtain moneyfrom the Nat~onal ScienceFoundationandpublishin scientificjour.nals. Many biologistshavealsofoundtheir viewsbiologicallyinteresting . If theirspeciesconceptis to berejected(asI think it must), it will haveto be on groundsotherthanthosethat Kitchermentions .

I. Monisfu, Pluralism, Unity and Diversity

What both sides of the controversy need to do is to state criteria for including or excluding particular definitions of the speciescategory and to give reasonswhy these criteria are appropriate. Ereshefsky(1992), Stanford (1995), Hull (1997), and Mayden (1997) have attempted to do just that. Ereshefskygroups the myriad definitions of the speciescategory into three basic types: interbreeding, ecological, and phylogenetic. He then evaluates each type according to two basic kinds of principles- sorting and motivating " " . Sortingprinciples sort the constituentsof a theory into basic units, and " " motivating principles justify the use of sorting principles (Ereshefsky1992, 682). With respect to the speciescategory, Ereshefskyprovides a single sorting principle and four motivating principles. The sorting principle for a " taxonomic approach should produce a single internally consistent taxon" omy. The motivating principlesare empirical testability, consistency with well-establishedhypothesesin other scientific disciplines, as well as consistency with and derivability from the tenets of the theory for which the taxonomy is produced (Ereshefsky1992, 682). When Ereshefskyapplieshis criteria to a variety of speciesdefinitions, he ' concludesthat the only respectabledefinitions are historical; none of Kitcher s structural definitions are up to snuff. The interbreeding, ecological, and phylogenetic definitions are acceptablebecausethey are basedon evolutionary theory (characterizinga process) and phylogeny (the product of this process). Ereshefsky(1992, 684) dismisses both the creationist and phenetic species also dismisses the typo logical species concepts, as one might expect, but he " concept of the idealistsbecauseit is incompatiblewith current evolutionary " biology . Kitcher would surely reply that there is more to biology than evolutionary biology . Perhapssomeother areaof biology , suchasdevelopmental biology , which is (largely?) independentof evolutionary biology might need a nonevolutionary way of grouping organisms, call them speciesor not. I feel required to add that the typo logical or Aristotelian speciesconceptis not quite as dead as Ereshefskymight lead one to believe. It is still alive and well in Catholic universities around the world (even after the current pope put his imprimatur on evolution); it plays a central role in the objectivist ); it raises its head once philosophy of Ayn Rand (a.k.a., Alisa Rosenbaum 1979, 1985); and Atran Platnick cladists with in connection ( pattern again of salient that (1990) argues species living creaturestypo logically perceiVing is built into our genes. We are all born typologists, like it or not. Finally, ' Ereshefskys requirementthat eachapproachmust produce a singleinternally 3 consistentclassificationmight strike somepluralists as too monistic. ' s brand of Stanford (1995) reacts primarily to Kitcher pluralism, arguing between for that it lacks adequate criteria acceptableand distinguishing ' s brand of realism Kitcher unacceptablespeciesconcepts. (He also argues that is incompatiblewith his pluralism, but that is another story .) Building on ' ~ tcher s book TheAdvancementof Science(1993), Stanford (1995) proposes to basehis criteria on the progressthat we makein the questionsthat we ask " in science. Speciesdivisions are the handmaidensof erotetic progress: They

Hull: On thePluralityof Species

enableus to makethesignificantquestions throughwhichwe extendsuccessful " 79 schemata moretractable in original). Thus, we canrejectspecies ( , emphasis definitionsif they areredundant , boring, or wrongheaded . A species division is redundantif it fails to "makeany significantquestionsmoretractable ." A " speciesdivision is boring if it doesnot help us to pursuefurther goals." Finally, a speciesdivision is wrongheadedif the schemataon which it is based"involvepresuppositions we believeareincorrect" (Stanford1995, 80). In his ensuingdiscussion , Stanfordclarifieshis criteriaby applyingthemto . Creationismis wrongheaded particularcases . The explanatoryschemata of " the creationists"restuponsubstantiallymistaken presuppositions(Stanford 1995, 80). For example , they attempt to argueaway the implicationsof carbondatingby postulatinga directionalchangein the rate of radioactive decay.They explainthe patternsto be foundin the fossilrecordin termsof which organismscould climb, swim, or fly the highest during the Great Flood. I needgo on no further. Accordingto Stanford is boring. , pheneticism ' The "naturaldependencies identifiedby the pheneticists OperationalTaxonomicUnits aretrivial andunhelpfulin pursuingany practicalend" (but see Dupre, chapter1 in this volume). The pheneticists would find this objection especiallydamningbecause they takethe practicalusefulnessof their classi ficationsto be one of their chiefvirtues. In defenseof the phenetic species , even Mayr (1981), the chief opponentof pheneticism concept , finds the establishment of phenaan importantfirst stepin the recognitionof genuine . species In the discussion of his examples , Stanfordmakesit clearthat applications of his criteriaarehistoricallycontingent . Rightnow, the creationistandpheneticspeciesconceptscanbe rejected , but in the pastthey might well have led to scientificprogress . For example , the appealof pheneticism restedon a common if , not universal , convictionthat somethingout thereexists quite that answersto the name"overallsimilarity." If Atran (1990) is right, this modeof perceptionmayhavea significantgeneticbasis.If nothingelse, the pheneticistsshowedthat, contrary to their own goals, no such thing as overallsimilarityexists. If suchbright, hardworking , and creativescientists in a period of twenty yearsor so could not comeup with anythingeven approximatingoverall similarity, it is very likely not to exist in the first ' place. In Cuviers day; theremight havesomethingto sayfor structuralism , but lessso today.

HULLAND MAYDENON SPECIFS In a recentpaper(Hull 1997) I usedthreephilosophicalcriteria- universality , applicabilityand theoreticalsignificance - to evaluatenine speciesconcepts currentlybeing.entertainedby professional biologiststo seeif any of . themscoremore on these criteria than do others . Thegoalwasto see highly if the very generalphilosophicalcriteriathat philosophershave developed to eval.uatescientificconceptscandistinguishbetweenthe nine speciescon-

I. Moni ~ , Pluralism , Unity and Diversity

cepts. To begin, scientists value the universalityof their concepts (but see Boyd, chapter 6 in this volume). For example, any de6nition of elementmust apply to all matter, not to just a subset. Physicistswould be lessthan pleased if their element concept applied only to metals or to nonradioactive substances . Biologists would like their preferredspeciesde6nition to apply to all organisms, not just some, but fulfilling this desideratumhas proven to be very difficult. For example, the biological speciesconcept applies to only those organismsthat reproducesexually, at least on occasion. . Biologists, like all scientists, would prefer that their conceptsbe applicable that the circumstances , way Perhapsthey neednot be totally applicablein all At the better. the are the more but , , they applicable operationists propose be never can a that in such least they applied , defining concepts way very runs counter to the testability criterion of science. The testing may be difficult , indirect, and fallible, but it must be possible. Philosophersare usually content once we have decidedthat a particular concept in scienceis in principle applicable, but scientistswant more, much more. They want grouping criteria- criteria that enable them to decided whether two or more organisms belong qr do not belong to a speciesin a significant percent of the cases. They also want ranking criteria- criteria that enable them to decide whether a taxon is a subspecies , species, or genus (Mishler and Brandon 1987). Finally, philosophers of scienceare currently convinced that to be useful, all scientific conceptsmust be theoreticallysignificant. They must function in a significant scientific theory. Becauseno scientific concept can be totally theory free, the issue becomes which theory colors which concepts (see Dupre, chapter 1, this volume). Next we must rank these theories according ' to how fundamentalthey are. To use the traditional example, Newton s theory of universal gravitation (once fixed up) is more fundamental than ' Kepler s laws of planetary motion (once fixed up). On this view, those concepts required by the most fundamental theories take precedenceto those is conceptsrequired by less fundamental theories. And if theory reduction the to reduced can be possible, all of these various upper level theories lower-level theories. With respect to the connection between process theories and the patterns discerniblein nature, many scientistsdisagreewith have philosophers about the primacy of theories. Perhaps philosophers have not. scientists They worked their way free of inductivism, but many insist that all scientific investigations must begin with direct, theory-free observation and proceed as cautiously as possible, avoiding idle speculation (seethe papersin Claridge, Dawah, and Wilson 1997). In my paper, I grouped the speciesconcepts that I evaluated into three families. The first family includes speciesconcepts that determine species statusin terms of some.form of similarity- for example, traditional morphotogical species concepts, the phenetic species concept, as well as certain , all of these speciesconcepts molecular concepts. Until the past few decades were typo logical in the sensethat a single list of dtaracters was developed

ofSpecies Hull:OnthePlurality

and only those organismsexhibiting all of thesecharactersare consideredto belong to the samespecies.Now , cluster analysisis the norm for estimating similarity (see Hull 1965 and Boyd, chapter 6 in this volume). The second family of speciesconcepts has been generatedby evolutionary biologists. The intent is to discern basic units of evolution. Included in this group are the evolutionary speciesconcept that can be traced back to Simpson(1961) ' and Hennig (1966), especiallyas developed by Wiley (1981); Mayr s (1969) ' ; Patersons (1981) speciesmate recognition concept biological speciesconcept ' s 1989 cohesion ) ; and Templeton ( concept. of The third family speciesconceptsoverlaps the preceding two families of concepts becauseI have distinguished it by means of historical considerations . I have constructed it in this way becauseof the historical connection ' to Hennig s phylogenetic systematics.4 Hennig (1966, 32) considered himself to be adopting a speciesconcept common in his day- something like the conceptsof Simpsonand Mayr (but seeMeier and Willmann 1999). When the reticulate relationships of sexual organismsare rent and splitting ' occurs, then one specieshas evolved into two species. However, Hennig s descendantshave developed species concepts that, though they may be descended&om his ideas, are not exactly coincident with them. The first is the monophyletic (or autapomorphic) species concept of Mishler and Donoghue (1982), Mishler and Brandon (1987), de Queiroz and Donoghue (1988, 1990), and Mishler and Theriot (1999). According to the mono phyletic speciesconcept, a speciesis the least inclusive monophyletic group definable by at least one autapomorphy. Hennig limited the application of the term monophylyto higher taxa. The advocates of the monophyletic speciesdefinition extend it to cover speciesas well. The diagnostic speciesconcept also grew out of Hennig and, in this case, Rosen (1978, 1979), but it has taken its own direction (Platnick 1977, Eldredge and Cracraft 1980, Nelson and Platnick 1981). The most influential formulation of the diagnostic speciesconceptwas first presentedby Cracraft (1983) and then further developed by McKitrick and Zink (1988), Nixon and Wheeler (1990), Wheeler and Nixon (1990), and Vrana and Wheeler (1992), among others. According to this view, a species is the smallest diagnosablecluster of individual organismswithin which there is a parental pattern of ancestry and descent. The monophyletic speciesconcept emphasizes phylogeny, though not the processes that give rise to phylogeny. The diagnostic speciesconcept acknowledgesthe importance of genealogy, ancestor-descendantrelations among organisms, but not necessarily phylogeny. It does not depend on the evolutionary process over and above genealogy. In the span of this essay, I cannot go through all of the particulars of the preceding speciesdefinitions and their variants. At the very least, the presence of so many speciesdefinitions taken seriouslyby professionalbiologists should warm the heartsof pluralists. As much as eachof theseauthors is sure that he has the correct speciesdefinition, such monistic inclinations have yet

/ I. Monism, Pluralism , Unity and Diversity

to narrow this list of speciesdefinitions significantly. As far as universalityis concerned, the phenetic speciesconcept is the most general conceptbecause it appliesequally to all organisms. The price that one pays for universality of this sort is that malesand femalesmay end up not being classedin the same species. The femalesof one speciesmay be more similar to the femalesin another speciesthan they are to the malesof their own species.On a strictly . phenetic speciesconcept, that eventuation would be perfectly acceptable because universal least are the The biological and mate recognition concepts they apply only to those organismsthat reproducesexually with reasonable frequency. Where these speciesconceptsapply, they distinguish significant evolutionary units. The trouble is that they do not apply to organismsthat existed during the first half of life on Earth or to many groups of organisms today. During the first half of life on Earth, evolution occurred, but not with the aid (or effect) of species(seeEreshefsky, Sterelny and Nanney, this volume ). All other speciesconceptsare arrayedbetween thesetwo extremes. As far as applicabilityis concerned, the speciesdefinitions that I treated are from the phenetic speciesat one end to the arrayedmuch more continuously . evolutionary speciesconcept at the other end. The phenetic speciesconcept is the most operational becausethat was the main reasonfor developing it . The diagnostic speciesconcept is the next most applicableconcept. Systematists must know who tends to mate with whom and what the results of theseunions are for sexualorganisms, as well as who gives rise to whom for asexualorganisms. The only other bit of information that they need is character covariation. The monophyletic speciesconcept is as operational as the methods of cladistic analysis permit. The goal of cladistic analysis is the individuation of charactersso that they nest perfectly. The mate recognition, biological, and cohesion concepts are even more difficult to apply because the forces and mechanismsthat they specify are more difficult to discern. Finally, the evolutionary speciesconcept is most difficult to apply becauseit explicitly specifiesthat speciesare extendedin time. Decisionsabout species status are contingent upon what will happen in the future. Speciescan be determinedonly in retrospect(seeSterelny, chapter 5 in this volume). , only the phenetic speciesconcept is With respectto theoreticalsignificance designed to be totally theory neutral or theory free. The diagnostic species concept assumesonly soine very low level, unproblematictheories, whereas all other speciesconceptsare openly theoretical in their content. Whether or not this characteristicis a virtue or a vice varies in the systematicscommunity . Someconceptsassumeknowledge of the evolutionary process. Others assumeonly phylogenies regardlessof the processes that produced them. " " S !l\e are nondimensional- that is, extended only minimally in spaceand time. Others treat speciesas lineages. When I first set myself the task of evaluating representativespeciesconcepts on the basisof widely assumedphilosophicalcriteria for good scientific classifications , I assumedthat one or two would emerge as better than the others. After all, I am a monist. However, no matter how I massagedthe

Hul\~On t' Pluralit\J of Species

data, I could not produce the result I had anticipated. All of the species conceptsI evaluatedscoredabout the same! One reasonfor this outcome is that the most easily applied conceptstend to be those with the least theoretical commitment, whereasthose conceptsthat produce theoretically significant speciestend to be the most difficult to apply. Universality, in its turn, does not covary with either theoretical significanceor easeof application . Some theoretically committed species concepts, such as the monophyletic speciesconcepts, apply to all organismsand are moderately easy to apply. The grudging conclusion of my paper (Hull 1997) is that none of the speciesconcepts that I evaluated are all that superior to the others- that is, if universality, applicability, and theoretical significance are weighted equally. However, in this samevolume, Mayden (1997) set himself the same task, but cameto quite a different conclusion. Mayden's goal was to find the primary speciesconcept. The differencesbetween our papers is instructive. First, Mayden recognizes twenty -two different species concepts, and he combines as single concepts several formulations that others take to be separateand distinct speciesconcepts. For example, the two formulations of the phylogenetic speciesconcept that I classifiedas two separatespecies conceptsare classifiedby Mayden as a single speciesconcept. Next , he evaluates these twenty -two conceptsaccording to their " convenience, accuracy, " precision, and the successfulrecovery of naturalbiological diversity (Mayden 1997, 381). Finally, Mayden evaluateshis speciesconceptson severaladditional criteria that are a good deal more specific. To serve as the primary speciesconcept , a concept must be theoretically significant and include sexual, asexual, and hybrid species ; it must be a nonrelational lineage concept that treats as individuals rather than as classes species ; and it must place no constraints on necessaryattributes that a speciesmust possessin order to be validated. As the primary speciesconcept, it need not be operational becauseother, secondary speciesconcepts provide the operational basis for this primary speciesconcept. For the primary speciesconcept, theoretical significanceis of primary importance. Only after a speciesconcept passesthis test do the other criteria come into play. The only speciesconcept that fulfills all of these criteria is the evolutionary species concept as reworked by Wiley (1981) and by Wiley and Mayden (1999). The main reasonwhy Mayden and I cameto very different conclusionsin evaluating various speciesdefinitions is that I combined theoretical significance and operationality. A speciesconcept might score quite highly on theoretical significance, but if it was not very operational, it endedup with a ' mediocrecumulative score. Mayden, to the contrary, took theoretical significance as necessaryand then rankedtheoretically significant speciesaccording to his other criteria, including operationality. A secondreasonwhy Mayden' s conclusion is so different Horn mine is that he included more substantive criteria. Yes, I think that any adequatespeciesconcept must treat speciesas

I. Mo

m, Pluralism , Unity and Diversity

lineages, but as a philosopher evaluating species concepts &om the outside , I felt committed to being open -minded about such issues. As a scientist engaged in these disputes , Mayden felt no such compunction . As a result , he was able to arrive at a single primary species concept .

CONCLUSION " " Postmodemistshave made positivists all-purpose whipping boys, usually parodying their views in the process. Other authorshave also joined in these parodies. For example, just about everyone claims that attempting to demarcate sciencefrom nonscienceor pseudoscienceis terribly wrongheaded, but ? Someof it is very bad science; then what do we do about creation science some of it is not scienceat all (Reisch 1998). Has philosophy really become ' so sophisticatedand nuancedthat we cant distinguish between scienceand ? My fellow philosophersare likely to respond that courts of creation science law are not graduate seminars, and they insist on limiting themselvesto is irrelevant. graduateseminars. The rest of ' society In deciding which speciesconceptsto take seriously, we seeminextricably caught up in the issue of what counts as genuine scienceand what not. Kitcher rejectsthe creationist and pheneticspeciesconceptsfor philosophical reasons. The problem with pheneticism is that it comes into conflict with his philosophical monism of the moment- namely, that no such things as theory-free observations, let alone concepts, exist. Hence, any attempt to deAne the species category in a theoretically neutral way is impossible. Kitcher is putting his bet on this philosophicalposition prevailing for awhile. ' I share Kitcher s prediction. Not only will philosophers continue to value theoretical significancehighly , but I am betting that an increasingnumbersof systematistswill come to sharethis conviction as phylogenetic cladists win out over their pattern cladist brethren. But theoretical significanceonly narrows the number of philosophically acceptablespecies concepts. For those of us who are more monistically inclined, traditional philosophical criteria alone are not sufficient forevaluating . speciesconcepts. On this score, thereis more to sciencethan philosophy exists Right now, an extremely powerful, well articulated theory actually in biology - evolutionary theory. This theory places constraints on both other speciesdefinitions and traditional philosophical desiderata. Of course, articulating in little a are headway theories are possible. Some scientists making alternative ways of viewing the living world , but until the promise of thesealternativesis realized, we cannot treat them on a par with evolutionary th.eory. In connection with our understanding of the evolutionary process, Sober (1984a, 335) is ilguessingthat [the] species-are-individuals perspective ' s conviction but , ':Villwinll (but seeWheeler and Platnick 1999). I shareSober .someday, way down the road, this perspectiveon speciesmay be overturned. The possibility of future alternatives is not, however, a sufficiently strong reasonfor acceptingthe pluralist philosophicalmonism of the moment.

/ Hull: On thePluralityof Species

Mayden (1997) concludes that one speciesconcept is preferable to all others for the role of the primary speciesconcept. None too surprisingly, the preferablespeciesconcept is the one that he prefers- the revised evolutionary ' speciesconcept. Isn t Mayden simply reasoningin a circle? He started his investigations preferring the revised evolutionary speciesconcept, and he concluded that it was preferable. However, even biased investigations can turn out differently from our expectations. I know that mine did. Mayden might have discovered to his dismay that the revised evolutionary species concept does not stand head and shouldersabove its competitors. Stranger things have happened. Mayden introduces criteria for evaluating species concepts that are closely connectedto the sort of sciencethat he wants to conduct. As a scientist engaged in the process that he is investigating, he cannot play it coy. He must commit himself. These additional commitments are what allow him to selectone speciesconceptas preferable. As far as strategies in scienceare concerned, sometimesscientists work themselvesinto a tight comer. They can seeone view of the world and one view only . During such times, I would join in the cry for pluralism. We need to get flexible, proliferate alternatives even if they are not very well supported , and so on. Sometimes , however, scientists are lost in conceptual brambles. Too many alternativespresent themselves, and there seemsto be no way to decideamong them. During thesetimes, a strong dose of monism is called for to help prune the tree of knowledge. With respectto the species problem right now, the situation clearly seems to exemplify the second extreme. Weare drowning in a seaof speciesconcepts. Hence, scientistsare justified in being more monistic than they have in the past. Perhapsmore than one speciesconceptis justified, but twenty -two? But what do I think? Did God createa single sort of speciesor many different sorts? Is there a single level of organization acrossall organismsthat is in some significant sensethe " same"1ff we take for granted the traditional organizational hierarchy of cells, organs, organisms, colonies, populations, and so on, the answer is clearly no. What if we take as the basic level the level at which reticulation is converted to divergence? This level is certainly significant as far as the evolutionary process is concerned, but in certain areasof the phylogenetic tree, organismsexhibit this level of organization, in other areas colonies exhibit this level of organization, in other areas , and in still other areashigher taxa. In asexualuniparental organisms species that do not exchangegeneticmaterialevenparasexually,splitting occursat the level of single organisms. Are we to call eachof these organismsa separate species(see Dupre, chapter 1, Nanney, chapter 4 in this volume)? In sexual , reticulation doesnot ceaseuntil speciationhas occurred, but in many . species groups of organisms, reticulation is relatively prevalent even among organisms commonly classedas belonging to different genera. The problem seems . to be the traditional organizationalhierarchy. As long as it seemsso right to us, any efforts to distinguish a single level of organization that counts as the sameacrossall organismsmay be very strongly counterintuitive. Perhapswe need to changeour intuitions.

/ I. Monism . Pluralism . Unity andDiversity

ACKNOWLEDGMENTS I wish to thank John Dupre , Marc Ereshefsky, Kim Sterelny , M . H . V . Van Regenmortal , Ed Wiley , Rob Wilson , and an anonymous referee for reading and commenting on this paper . NOTFS 1. As Dupre remarkedin responseto the examplegiven, alwaysusing virus-impenetrable of the rubberswoulddefeatthe AIDSepidemicin two ways: it wouldpreventthe transmission would also the but in the it virus to new hostsvia sexualintercourse , process preclude birth suchprophylacticswould alsobe spermimpenetrable of new humanbeingsbecause , a cure . decidedlyworsethanthedisease 2. Kitcher(1987, 187) enlistsMishlerandDonoghue(1982) asfellow pluralists , but they protest . "Kitcher'S (1984a , 1984b ) brandof pluralismimpliesthat thereare many possibleand complex for a given situation(say, the Drosophila mtllmogaster pennissiblespeciesclassi6cations . In contrast , Mishlerand ), dependingon the needsandinterestof particularsystematists ' -purposeclassi6cation Donoghues (1982) brandof pluralismimpliesthata singleoptimalgeneral . in each situation the criteria existsfor eachparticularsituation , but that may well be applied ton (1989) a Kitcher-stylepluralist.He " (MishlerandBrandon1987,403). Nor is Temple different - cohesiveness andonecharaderonly is relevantto speciesstatus thinksthat onecharacter cancontributeto thiscohesiveness eventhoughdifferentmechanisms (seealsoDonoghue1985 andEreshefsky 1992). ' . Because 3. Pluralismis at bottom incompatiblewith WheweUs conswenceof inductions a surprise it comes as of this MichaelRuseis amongthe mostenthusiastic , prindple supporters to find him on Ereshefsky , albeitas highly conservative ' S list of pluralists pluralist. Although into groups , and Ruse(1987, 238) arguesthat thereare"differentwaysof breakingorganisms on thispoint. " I happento thinkthathe is far too sanguine theycoinddtf mirror the decisions 4. I amawarethat the dedsionsI havemadeat this higherlevelof abstraction to considerations . historical othershavemadeat thelevelof species My allowing concepts overridesimilaritywith respecito how I classifyspecies conceptsis likely to imply something level. I find mostimportantat the species aboutwhichconsiderations

REFEREN CFS . Cambridge : Towards ananthropology of science Atran, S. (1990). Cognitive of nRturlllhistory foundations . : Cambridge UniversityPress 241, 515. AIDS. Science Blattner , W., R. C. Gallo, andH. M. Temin(1988). HIV causes . : Theunitsof biodiversity , andMR . Wilson. eds. (1997). Species daridge, M. F., H. A Dawah andHall. London:Chapman . SocialStudies of relativism Collins, H. M. (1981a of Science ). Stagesin the empiricalprogramme 11, 3- 10. . : The soda ! destructionof a physicalphenomenon Collins, H. M. (1981b ). Sonof sevensexes 11, 33.- 62. SocialStudies of Science . . asa methodological .Collins,H. M. (1981c ). Whatis TRASPl:Theradicalprogramme imperative 11, 215- 224. of theSocialSciences Philosophy . In R. F. Johnson . ed., CurrentOrnithol Cracraft , J.-(1983). Species analysis conceptsandspecies Press . : Plenum . New York ogy

/ Hull: On the Plurality of Species

de Queiroz , K , andM. J. Donoghue(1988). Phylogenetic systematicsand the speciesproblem. Cladistics 4, 317- 338. de QueiroZ, K. , and M . J. Donoghue (1990). Phylogenetic systematicsand speciesrevisited. Cladistics6, 83- 90. de Sousa . R. (forthcoming) . Biological individuality . Biologyand Philosophy . Dick, S. J. (1982). Plurality of worlds: Theoriginsof the utra terrestriallife debatefrom Democrltusto Kant. Cambridge: CambridgeUniversity Press.

, M. J. (1985). A critiqueof the BSCandrecommendations for a phylogeneticalternative Donoghue . TheBryologist 83, 172- 181. as carcinogens and pathogens : Expectations Duesberg , P. H. (1987). Retroviruses and reality. Cancer Research 47, 1199- 1220. theAIDSI1irus . WashingtonD.C.: Regnery . Duesberg , P. H. (1996). [nl1enting Review 90, 66- 90. Dupre,J. (1981). Naturalkindsandbiologicaltaxa. Philosophical : Metaphysical Dupre, J. (1993). The disorderof things . foundations of the disunityof science , Mass.: HarvardUniversityPress . Cambridge Eldredge . Oxford: OxfordUniversityPress , N. (1985). Unfinished . synthesis Eldredge , N., andJ. Cracraft(1980). Phylogenetic andtheeoolutionary : Methodand patterns process . NewYork: ColumbiaUniversityPress . theoryin comparatioe biology . Philosophy , M. (1992). Eliminativepluralism 59, 671- 690. Ereshefsky of Science Fisher , R. A. (1958). Lungcancerandcigarettes ?Nature182, 108. Ghiselin , M. T. (1974). A radicalsolutionto thespecies . Systematic 23, 536- 544. problem Zoology Ghiselin , M T. (1989). Sexand the individualityof species : A reply to Mishlerand Brandon . andPhilosophy 4, 73- 76. Biology -Merrill. andprojects Goodman , N. (1972). Problems . Indianapolis : Bobbs Griffiths, P. E., andR. D. Gray (1994). Developmental . systemsandevolutionaryexplanations 91, 277- 304. Journal of Philosophy . Chicago : Universityof DlinoisPress . Hennig , W. (1966). Phylogenetic systematics , K. E. (1984). Thenatureof biologicalspecies . Philosophy 51, 293- 307. Holsinger of Science : Pluralismandspecies , K. E. (1987 ). Discussion Holsinger , or whenmustwe agreewith concepts oneanother ? Philosophy 54, 480- 485. of Science HulL D. L (1965). The effectsof essentialism on taxonomy . BritishJournal for thePhilosophy of Science 15, 314- 326, and16, 1- 18. Hull, D. L. (1976). Are species ?Systematic 25, 174- 191. reallyindividuals Zoology HulLD. L. (1978). A matterof individuality.Philosophy 45, 335- 360. of Science HulL D. L. (1997) . The ideal speciesdefinition and why we can't get it . In M . F. daridge , H. A : Theunitsof biodioersity Dawah, and MR . Wilson, eds., Species . London: O1apmanand Hall. . Kitcher, P. (1984a). . Philosophyof Science 51, 308- 333. Species Kitcher, P. (1984b). Ag ~ Science 51, 616- 630.

t the monism of the moment: A reply to Elliott Sober. Philosophyof

Kitcher, P~(1987). Ghostly whispers: Mayr , Ghiselin and " the philosophers" on the ontology of species.Biologyand Philosophy2, 184- 192.

I. M~ sm , Pluralism , Unity andDiversity

. In M. Ruse Kitcher,P. (1989). Somepuzzlesaboutspecies , ed., WhRtphilosophy of biologyis not. . : Dordrecht Kluwer . . NewYork: OxfordUniversityPress Kitcher,P. (1993). TheIIdvancement of science . Journalof ' s monism Kitcher, P., K. Sterelny , andK. Waters(1990). The illusoryrichesof Sober 161 . 87 158 , Philosophy in the sagaof the : The denouement , R. L (1997 ) . A hierarchyof speciesconcepts Mayden : Theunitsof . Wilson . In M. F. Claridge , eds., Species , andMR , H. A Dawah speciesproblem andHall. . London:Chapman biodiversity . Nature393, 634- 640. MaynardSmith,J. (1998). Theoriginof altruism . NewYork: McGraw-Hill. of systematic zoology Mayr, E. (1969). Principles . Science of opposingmethodologies : Towarda synthesis Mayr, E. (1981). Biologicalclassification 2'14, 510- 516. . In Q. D. Wheelerand R. Meier, eds., Species Mayr, E. (1999). The biologicalspeciesconcept . : ColumbiaUniversityPress : . New York A debate and theory concepts phylogenetic 90, 1- 14. in ornithology. Condor McKitrick , M. C., andR. M. Zink (1988). Species concepts . In Q. D. WheelerandR. Meier, Meier, R., andR. Willmann(1999). ~ e Hennigianspecies concept . : A debate . New York: ColumbiaUniversityPress andphylogenetic eds., Species theory concepts , andthe phylogenetic Mishler, B. D., andR. N. Brandon(1987). Individualism , pluralism species 2, 397- 414. . BiologyandPhilosophy concept . Systematic : A casefor pluralism Mishler, B. D., andM. J. Donoghue(1982). Species concepts . 503 31 491 , Zoology , andphylogeneticspeciesconcepts Mishler, B. D., andE. Theriot(1999). Monophyly, apomorphy : A debate . New andphylogenetic . In Q. D. WheelerandR. Meier, eds., Species theory concepts . York: ColumbiaUniversityPress . In D. Hull, M Forbes , Moss, L (1992). A kernelof truth? On the realityof thegeneticprogram of Science Mich.: Association , 1. East 1992 vol. eds . PSA andK. Okruhlic , , , Philosophy Lansing , 335- 348. . New andvicariance : Cladistics andbiogeography Nelson , G., andN. Platnick(1981). Systematics . York: ColumbiaUniversityPress . Nixon, K. C., andQ. D. Wheeler(1990). An ampliAcationof the phylogenetic speciesconcept 223 . 6 211 Cladistics, : A critiqueof Paterson , H. E. H. (1981). Thecontinuingsearchfor theunknownandunknowable - 119. 77 113 Science . South , ideas on Journal of African speciation contemporary , andhypothesistesting. Systematic Zoology , phylogenetictrees Platnick , N. (1977). Cladograms 26, 438- 442. . Systematic of cladistics Zoology28, Platnick , N. (1979). Philosophyand the transformation 537- 546. 1, 87. Cladistics of cladisticsrevisited Platnick , N. (1985). Philosophyand the transformation ' 94. . Philosophy , andthe problemof pseudoscience Reisch , logicalempiricism , G. A (1998). Pluralism . ' Sdence of . Systematic Rosen , D. E. (1978). Vicariantpatternsand historicalexplanationin biogeography 27, .159- 188. Zoology

/ Hull: On thePluralityof Spedes

basinsof guatemala . Bulletinof Rosen , D. E. (1919). FishesHornthe uplandsandintermontane theAmerlCRnMuseumof NaturalHistory162, 261- 316. in the languageof ! dence . Dialogue14, Rosenberg , A (1975). The virtues of vagueness 1.81- 305. Ruse : Naturalkinds, individuals , M. (1987 ). Biologicalspecies , or what? British] oumalfor the 38, 1.1.5- 1.41.. of Science Philosophy . Ada BiotheoretiC R 45, Shanahan . T. (1997 , antirealism , and the units of selection ) . Pluralism 117- 11 . 6. . New York: Columbia University Press. Simpson, G. G. (1961). Principlesof animal ta. ronomy Sober, E. (19848). Discussion: Sets, speaes, and evolution. Comments on Philip Kitdter' s " ." Philosophyof Science Sl , 334- 341. Species Sober, E. (19Mb ). Thenatureof selection . Cambridge, Mass.: MIT Press. Sober, E. (1993). Philosophyof biology. Boulder, Colo.: Westview Press. Sober, E., and D. S. Wilson. (1994). A aitical review of philosophical work on the units of selec-

tion problem . Philosophy 61, 534- 555. of Science

Sober : theevolution andps.vchoiog behaoiors . , E., andD. S. Wilson. (1~ 8). Untoothers .vof unselfish : HarvardUniversityPress . Cambridge 62, Philosophyof Science

85, 339- 361. , K. , andP. l 0, which satis6esthe following condition: (Vx, ye T) [if x K y then R (y) > R (x)]. The integers m, . . . , n in the range R are called rankswith respect to R, and R(x ) is the rank of x with respectto R. 1.2.2.1. Rank n is the rank of folk kingdom(FK). 1.2.2.2. Rank 0 is the rank of genericspecies (GS). 1.2.2.3. Rank n - 1 is the rank of life fonn (LF). 1.2.2.4. Rank - 1 is the rank of folk specific(FS). 1.2.2.5. Rank - 2 is the rank of folk varietal(FV). 1.2.2.6. Rank - 3 is the rank of folk subvarietal . 1.2.2.7. Taxa (named or unnamed) between ranks n - 1 and 0 are intennediate . 1.2.3. In any system of folk concepts, FK and GS' (i.e., ranks 0 and n) are mandatory in the sensethat every terminal kind is a subkind of sometaxon of that rank, such that 1.2.3.1. A terminalkind has no subkinds (i.e., x is terminal for K if and only if x is in the domain of K and there is no y such that y K z).

254

IV. ~ edesin Mind andCulture

1.2.3.2. VI [I is terminal- + (R (I ) = i V 3Y(IK y and R (y) = i ))] . 1.2.3.2.1. It follows that if T is a taxonomic category, the maximal rank n of the head of T is mandatory . 1.2.3.2.2. It also follows that if a level is mandatory, it partitions the taxa at that level or lower (into mutually exclusive groups of organisms ). 1.2.4. It remainsan open question, whether or not 1.2.4.1. LFs are mandatory. If so, then apparently unaffiliated generic speciesare in fact monotypic life forms; that is, the LF and its single GS are extensionally (perceptually ) equivalent, but conceptuallydistinct. 1.2.4.2. Someintermediatetaxa are ranked. If so, then any such intermediatetaxon is a subkind of somelife form, such = that: (VIE T )[R (I ) = n 2 > 3Y(IK y and R (y) n - 1)] . 1.2.5. In the historical developmentof Western systematics: 1.2.5.1. Rank n became the biological kingdom (Cesalpino 1583). 1.2.5.2. Rank 0 fissioned into ranks 0, the biological species, and 1, the genus (including monospecific genera) (T oumefort 1694). 1.2.5.3. Rank n - 1 became the biological class (Linnaeus 1738). 1.2.5.4. Rank n - 2 was initially formalized as the biological family (Jussieu1789). 1.2.5.5. Taxa below rank 0 became unranked infraspecific groups (Darwin 1859). FB is domain specific(rather than domain general) in the sense that it involves cognitive operationspartially dedicatedto the perceptualidentification and conceptualprocessingof nonhuman living kinds as generic speciesof animalsand plants: 2.1. Experimentswith people in highly diverse cultures (U.S. Midwest, Mayan rainforest), suggestthat 2.1.1. People everywhere categorize and reason about nonhuman and ranked ~ organismsin terms of essencebasedgenericspecies taxonomies. . 2.1.2. Domain-general, similarity-based models fail to account for thesefindings.

255

J Taxonomy Atran: The' Universal Primacy of GenericSpeciesin Folkbiologica

This domain-specific capacity for the universal and spontaneousunderstanding of the organic world is 3.1. likely grounded in a priori abilities that are independent of actual experienceand thus innateto the human mind/ brain. 3.2. Plausibly an evolutionaryadaptation (rather than a by -product or accident) to relevant and recurrent features of hominid ancestral environments (e.g ., wide-ranging subsistenceinvolving the understanding of potentially indefinitely many speciesand habitats). Systematicbiology culturally evolved from folkbiology . Within an everexpanding global (and ultimately cosmic) framework, natural historians from Aristotle to Darwin sought ways to extend or overcome the limits of understanding inherent in people's spontaneousabilities to parse the local environment' s biodiversity .

ACKNOWLEDGMENTS The comparativestudiesreported here were funded by the National Science Foundation (SBR 93- 19798, 97-07761) and the FrenchMinistry of Research and Education (Contract CNRS 92-C-O758). They were codirected with Douglas Medin. Participants in this project on biological understanding across cultures include Alejandro Lopez (psychology, Max Planck), John Coley (psychology, Northeastern University . ), ElizabethLynch (psychology, Northwestern University), Ximena Lois (linguistics, Crea-Ecole Polytechnique ), Valentina Vapnarsky (anthropology, Universite de Paris X) , Edward Smith and Paul Estin (psychology, University of Michigan), and Brian Smith ( biology, University of Texas, Arlington ).

NOTFS 1. Thus, comparing constellationsin the cosmologiesof ancient China. Greece, and the Aztec Empire shows little commonality. By contrast, herbalslike the ancient Otinese ERH YA, Theo' phrastuss PeriPutonIstorias, and the Aztec BadianusCoda, shareimportant features, suchas the classmcationof generic speciesinto tree and herb life forms (Atran 1990, 276). 2. Generalizationsacross.taxa of the samerank thus differ in logical type from generalizations that apply to this or that taxon. Tentaite , pig, and lemonfreeare not related to one another by a simple classinclusion under a common hierarchicalnode, but by dint of their common rankin this case, the level of generic species. A system of rank is not simply a hierarchy, as some suggest ( Rosch1975, Premack1995, Carey 1996). Hierarchy- that is, a structure of inclusive - is common to many cognitive domains, classes including the domain of artifacts. For example, chair often falls under furniture but not vehicle , and car falls under vehiclebut not furniture. But there is no ranked systemof artifacts: no inferential link or inductive framework spansboth chair and car or fu~ iture and vehicleby dint of a common rank, such as the artifact species or the artifact family. 3. For example, in a comparativestudy of Itzaj Maya and rural Michigan college students, we found that the great majority of mammal taxa in both cultures correspond to scientific species and that most also correspondto monospecificgenera: 30 of 40 (75%) basic Mid1igan mammal

256

IV. ~

esin Mind andCul~

terms denote biological species ; , of which 21 (70%, or 53% of the total) are monospedficgenera 36 of 42 (86%) basicItzaj mammalterms denote biological species, of which 25 (69%, or 60 %of the total) are monospedfic genera ( U>pez et al. 1997, Atran 1999). Similarly, a Guatemalan government inventory of the Itzaj area of the peten rainforest indicates that 69 % (158 of 229) ' are monospecific(AHG / APESA 1992; d . Atran and Ucan Ek forthcoming), the samepercentage of monospecifictree genera(40 of 58) as in our study of the Chicago area( Medinet al. 1997) . 4. Moving vertically within eachgraph correspondsto changing the premisewhile holding the conclusioncategory constant. This maneuverallows us to test another domain-generalmodel of category-basedreasoning: the similarity- coveragemodel (Osherson et al. 1990). According to this model, the closer the premise category is to the conclusion category, the stronger the induction should be. Our results show only weak evidence for this general reasoning heuristic, " " which fails to accountfor the various jumps in inductive strength that indicate absoluteor relative preference. 5. By contrast, a partitioning of artifacts (including those of organic origin, suchas foods) is neither mutually exclusivenor inherent: somemugs mayor may not be cups; an avocadomay be a fruit or vegetabledependingupon how it is served; a given object may be a bar stool or a waste bin dependingon the social context or perceptualorientation of its user; and so on. 6. Although there may a cognitive susceptibility to radsm, the resultant social groups are by no meansas evident acrosscultures as are notions of generic species. Neither is the intermittent ranking of social formations (e.g ., armies) consistent across cultures or necessarilybound to essentializedgroups. In short, the apparent features of folkbiological taxonomy found in the domain of persons and social groups appearsto involve a variable transferenceof prindples from folkbiology rather than a common ontological foundation or mode of construal. The Rip side of this ontological transferenceis anthropomorphism- that is, the projection of human intentionality onto animals. Given the initial absenceof causalknowledge about (usually furtive) animals, the use of person analogies (especially by children) may initially lead to useful and accurate predictions about entities phylogenetically similar to humans (lnagaki and Hatano 1991).

CFS REFEREN RriofortStilldelDtpIlrillmtntodel dePet(n: Inr1tnt AHG/ APESA(1992). Plandedesarollo integrado . : SEGEPLAN Altmllnilly GUIlttmllla). SantaElena Gobiernos , Peten Pet(n (Conr1tnio . . Hillsdale CMrlldtr of thought Anderson , N.J.: Erlbaum , J. (1990). 11ItIldIlptir1e 87, 298- 315. Atran, S. (1985). Thenatureof folk botanicallife forms.AmtriCRnAnthropologist . Journalof theHistoryof Biology20, andgenusconcepts Atran, S. (1987 ) . Originsof the species 195- 279. . Cam: Towards an anthropology Atran, S. (1990). Cognitive of science of naturalhistory foundations bridge, England: CambridgeUniversity Press. Atran, S. (1993). Itza Maya tropical agro- forestry. CurrentAnthropology34, 633- 700. Atran, S. (1995). Causalconstraintson categoriesand categoricalconstraintson biological reasoning acrossculture. In D. Sperber, D. Premack . eds., Causalcognition.Oxford : . and A Premack Oxford University Press. 21, Atran, S. (1998b). Folkbiology and the anthropology of sdence. Behavioraland Brain Sciences 547- 611. . Atran, S. (1999). Itzaj Maya folkbiological taxonomy. In D. Medin and S. Atran, eds., Folkbiology MIT Press . Mass.: , Cambridge

257

Atran: The Universal Primacy of GenericSpeciesin Folkbiological Taxonomy

Atran, S., andD. Medin(1997 ). Knowledgeandaction: Culturalmodelsof natureandresource In M. Bazerman in Mesoamerica . , D. Messick , A Tinbrunsel , and K. Waydemanagement . . SanFrancisco : New LexingtonPress Benzoni , ethics , andbehavior , eds., Environment : Its placein culture.In L TolchinAtran, S., andD. Sperber(1991). Learningwithout teaching . Norwood andpsychological . N.J.: Ablex. , schooling , ed., Culture development sky-Landsmann ' of usefulplantsamongthe Northern Atran, S., and E. UcanEk (forthcoming ). Classification . PetenMaya(ltzaj). In C. White, ed., AncientMayadiet. SaltLakeCity: Universityof UtahPress andbasiclevels:Essence and Atran, S., P. Estin, J. Coley, andD. Medin(1997 ) . Genericspecies 17 22 45. in . , appearancefolkbiologyJournalof Ethnobiology Club Bartlett,H. (1940). Historyof thegenericconceptin botany. BulleHnof theTorreyBotanical 47, 319- 362. . In E. RoschandBUoyd, eds., Cogni Honandcate classi6cation Berlin , B. (1978). Ethnobiological . Hillsdale . , N.J.: Erlbaum gorimHon : PrincetonUniversityPress . Hon.Princeton Berlin, B. (1992). Ethnobiological classifica ' In D. and Medin S. Atran, eds., Berlin , B. (1999). One Maya Indians view of the plant world, MIT Press . . Mass . : , Folkbiology Cambridge andnomenclature Berlin, B., D. Breedlove , andP. Raven(1973). Generalprinciplesof classi6cation 74, 214- 242. in folkbiology.American Anthropologist . New York: Hon , andP. Raven(1974). Principles Berlin , B., D. Breedlove of Tzeltalplantclassifica Academic Press . . System aHc Bock , W. (1973). Philosophicalfoundationsof classicalevolutionarytaxonomy 22, 275- 392. Zoology Hon16, . Memoryand Cogni Boster , J. (1988). Naturalsourcesof internalcategorystructure 258- 270. . In Boster , J. (1991). The informationeconomymodelappliedto biologicalsimilarityjudgment Hon . WashingtonD Hveson sociall L Resnick , eds., Perspec , andS. Teasley , J. Levine V sharedcogni Association . .C.: AmericanPsychological . New andnaming : Unifonni Hesin folk classification andlivingthings Brown , C. (1984). Language Press . N .: Bnmswick , .J RutgersUniversity ? In J. PouillonandP. Maranda Bulmer , , R. (1970). Whichcame6rst, the chickenor the egg-head : Mouton. Levi-Strauss . TheHague : Melanges Hons eds., Echanges etcommunica offerisa Claude 13, 9- 28. . SocialScience Bulmer , R. (1974). Folkbiologyin theNew GuineaHighlands Information 5, 97- 109. . S.vstematic Cain , A (1956). Thegenusin evolutionarytaxonomy Zoology . in childhood . Cambridge , Mass.: MIT Press change Carey,S. (1985). Conceptual , eds., Carey, S. (1996). Cognitivedomainsasmodesof thought. In D. OlsonandN. Torrance . . NewYork: Cambridge Modesof thought UniversityPress : Marescot . , A (1583). DeplanHslibri XVI. Florence Cesalpino in ? Inductiveinferences ). Doesrankhaveits privilege Coley, J., D. Medin, andS. Atran (1997 112 . . 63 73 taxonomies , Cognition folkbiological E. . Proffitt . , D. Medin, and S. Atran (1999). Inductivereasoningin folk, J Coley, J, Lynch . . Cambridge D. Medin andS. Atran, eds., Folkbiology . In , Mass.: MIT Press biologicalthought . London : Murray. Darwin, C. (1859). On theoriginsof species of naturalselection bymeans

258

/ N . Speciesin Mind and Culture

., reprint1883). On theoriginsof species Darwin , C. (1811 o/ nRturalselection , 6th ed. New by means York: Appleton. Darwin , C. (1960- 1967). Darwin's notebookson b-ansmutationof species . G. de Beer, M. Rowlands , and B. Skarmovsky , eds., Bulletinof theBritishMuseum( NaturalHistory ) 2, 27200, and3, 129- 176(notebooksB, C, D, E) . - ornithologyof theKetengban Diamond , J., andD. Bishop(1998). Ethno New , Indonesian people Guinea . In D. MedinandS. Atran. eds., Folkbiology . Cambridge . , Mass.: tvfiT Press Diver, C. (1940). Theproblemof closelyrelatedspeciies . In J. Huxley, ed., living in thesamearea Thenewsystematics . Oxford: Clarendon Press . Donnellan. K. (1971). Necessityandaiteria. In J. Rosenberg andC. Travis, eds., Readings in the -Hall. . Englewood - aiffs, N.J.: Prentice of language philosophy andrelativityin classification . Amm ' canEthnologist , J. (1978). Salience 5, 66- 80. Dougherty , J. (1979). Learningnamesfor plantsandplantsfor names . Anthropological Dougherty Unguistics 21, 298- 315. . Cambridge . , Mass.: HarvardUniversityPress of things Dupre,J. (1993). Thedisorder Eldredge , N. (1986). Information , economicsand evolution. Annual Reviewof Ecologyand 17, 351- 369. Svstematics . Cambridge Ellen , R. (1993). Theculturalrelations : Cambridge . of classification UniversityPress . Philosophy 59, 671- 690. , M. (1992). Eliminativepluralism Ereshefsky of Science . Cognition Gelman , S., andH. Wellman(1991). Insidesandessences 38, 214- 244. beliefsin dtiIdren. in L. Hirschfeldand Gelman . S., J. Coley, andG. Gottfried(1994). Essentialist S. Gelman . , eds., Mappingthemind. NewYork: Cambridge UniversityPress -Merrill. .). Problems andprojects . Indianapolis : Bobbs Goodman , N. (1911 GreeneE. (1983). Landmarks in botany : StanfordUniversityPress(originally , 2 vols. Stanford in 1909). compiledfor theSmithsonian , D. Premack , and A Hacking , I. (1995). The lopping effectsof humankinds. In D. Sperber . Oxford: OxfordUniversityPress . Premack , eds., Causal cognition ' s naive Hatano 50, 171, G., andK. lnagaki(1994). Yo\U\g children theoryof biology. Cognition 188. Hatano , G., and K. inagaki(1998). A developmental perspectiveon informalbiology. In D. MedinandS. Atran. eds., Folkbiology . Cambridge . , Mass.: tvfiT Press classification and Hays, T. (1983). Ndumbafolkbiologyandgeneralprinciplesof ethnobotanical nomenclature . American 85, 592- 611. Anthropoiogist of anearlyconception Hickling , A , andS. Gelman(1995). How doesyour gardengrow? Evidence of plantsasbiologicalkinds. ChildDevelopment 66, 856- 876. de6nitionandwhy we can't get it. In M. Claridge Hull, D. (1997). Theidealspecies , H. Dawah , andM Wilson, eds., Species : Theunitsof biodiversity . London:Chapman andHill.

259

.

/ Atran: The Univenal Primacyof GenericSpeaesin Folkbiological Taxonomy

' , K., andG. Hatano(1991). Constrained lnagaD personanalogyin youngchildrens biological inference . Cognitive 6, 219- 231. development ' , K. , and G. Hatano(forthcoming ). Yo\ mg childrens recognitionof commonalities lnagaD betweenplantsandanimals . ChildDevelopment . um. Paris : Herissant . , A-L. (1789). Genera Jussieu plantar Keil, F. (1979). Semantic andconceptual : An ontologicalperspective . Cambridge , Mass.: development HarvardUniversityPress . Keil, F. (1989). Concepts . Cambridge . , kinds , andcognitive , Mass.: MIT Press development of naturalkinds. In D. Sperber Keil, F. (1995). Thegrowthof causalunderstandings , D. Premack , andA Premack . Oxford: OxfordUniversityPress . , eds., Causal cognition Kitcher,P. (1993). Theadvancement . NewYork: OxfordUniversityPress . of science Lassaline . Wisniewski , ME . andD. Medin (1992). Basiclevelsin artificialandnaturalcategories . In B. Burns andcategories . NewYork: Elsevier . , ed., Percepts , concepts Levi-Strauss . Journalof theRoyalAnthropological Institute93, , C. (1963). Thebearandthebarber 1- 11. Linnaeus : Wishoff. , C. (1738). Classes plantarum. Leiden Linnaeus botanica . Stockholm : G. Kiesewetter . , C. (1751). Philosophia Ll>pez, A , S. Atran. J. Coley, D. Medin, andE. Smith(1997 of folk). Thetreeof life: Universals . Cognitive andinductions 32, 251- 295. Psychology biologicaltaxonomies Mandler the sheepfrom the goats: Differentiating , J., P. Bauer , andL. McDonough(1991). Separating . Cognitive 23, 263- 298. Psychology globalcategories ' Mandler , J., andL McDonough(1996). Drinkinganddrivingdont mix: Inductivegeneralization in infancy.Cognition 59, 307- 335. . NewYork: McGraw-Hill. Mayr, E. (1969). Principles of systematic zoology . Cambridge . , Mass.: HarvardUniversityPress Mayr, E. (1982). Thegrowthof biological thought Medin, D., E. Lynch and reasoningamongtree . J. Coley, and S. Atran (1997 ). Categorization : Do all roadsleadto Rome ? Cognitive 32, 49- 96. Psychology experts -basedinduction . Osherson . D., E. Smith, o . Wilkie, ALl >pez, and E. ShaAr(1990). Category Review97, 85- 200. Psychological in naivebiology. In D. Sperber Premack , D. (1995). Forwardto partIV: Causalunderstanding , D. . Oxford: OxfordUniversityPress . Premack , andA Premack , eds., Causal cognition . , Mass.: HarvardUniversityPress Quine, W. (1960). Wordandobjed.Cambridge . In T. Moore, andsemanticcategories Rosch , E. (1973). On the internalstructureof perceptual ed., Cognitive andtheacquisition . NewYork: Academic . of language development andculturalspecifics in categorization . In R. Brislin, S. Bochner , and Rosch , E. (1975). Universals W. Lonner -culturalperspectives onlearning . NewYork: Halstead . , eds., Cross : Studiesin the internalstructureof natural Rosch , E., andC. Mervis(1975). Familyresemblances . Cognitive 8, 382- 439. Psychology categories -Braem(1976). Basicobjectsin natural Rosch , E., C. Mervis, W. Grey, D. Johnson . andP. Boyes . 8 382 439 . , Psychology categoriesCognitive Simmons of biological , D., andF. Keil (1995). An abstractto concreteshift in the development 56, 129- 163. thought: Theinsidesstory. Cognition

260

/ N . Speciesin Mind and Culture

. New York: ColumbiaUniversityPress . , G. (1961). Principles of animaltR% onomy Simpson andconcepts . Cambridge Smith . E., andD. Medin (1981). Categories , Mass.: HarvardUniversity Press . Sober . Boulder . , E. (1993). Philosophy , Colo.: WestviewPress of biology Sommers tree. Mind 68, 160- 185. , F. (1959). Theordinarylanguage : A naturalistic . Oxford: Blackwell . , D. (1996). &plainingculture Sperber approach of biologicallyspeaRcbeliefs : Thecaseof , K. , andF. Keil (1989). On thedevelopment Springer inheritance . ChildDlfJeiopment 60, 637- 648. ' " Stevens . " Systematic dassification , P. (1994). Berlins Ethnobiological Biology43, 293- 295. Stross . In M. Edmonson , B. (1973). Acquisitionof botanicaltenninologyby Tzeltalchildren , ed., . TheHague : Mouton. Meaningin Mayanlanguages andexpertise : Is thebasiclevelin the eyeof Tanaka , J., andM. Taylor(1991). Objectcategories thebeholder ?Cognitioe 23, 457- 482. Psychology Toumefort debotanique . Paris:ImprimerieRoyale . , J. (1694). Eltmens Wallace : Macmillan(1sted. 1889). , A (1901). DanDinism , 3rd ed. London Warburton of classi6cation . Systematic 16, 241- 245. , F. (1967 ). Thepurposes Zoology . In C. Craig Zubin, D., andK.-M. Kopcke(1986). Genderandfolk taxonomy , ed., Nounclasses andcategorization . Amsterdam : JohnBenjalnins .

261

/ Atran: The Universal Primacyof GenericSpeciesin Folkbiological Taxonomy

1 0

Species, Stuff , and Patterns of Causation

FrankC. KeilandDanielC. Richard son

Adults acrossall cultures seemto think about biological phenomenain distinct ways that are not found in other domains of knowledge. Thought about the living world has its own particular nature that may be different from thought about artifacts and the nonliving natural world . Perhapsthe most fundamentalunit of that apparently specialkind of biological thought is the concept of the species: the basic ontological kind around which so much of folk biology seemsto be organized. If this is correct, it is in many ways quite extraordinary, for the phenomenaldifferencesbetweenplants and animalsseemso enormousthat it is remarkablethat they nonethelessmight be conceivedas vastly more similar to eachother than either is to other sorts of kinds. In this essay, we ask how it is that speciesmight be thought of so differently and on what psychologicalbasis. At the most general level, two , nonmutually exclusive possibilities might explain this phenomenon: (1) there is something in the information about biological speciesthat is structured so differently that it shapeslearning differently and results in different kinds of cognitive structures for thinking about species ; and (2) humans have certain intrinsic cognitive blases that lead them to think about speciesin very different ways. The task here is to explore the details of such possibilities and their relative roles. To frame the problem, it is useful to seehow eachpossibility could be the only account. For example, it might be that the informational patterns associatedwith biological kinds result in different kinds of knowledge structuresbeing created by completely gener"a1learningcapacitiesin humans. The kinds of features seen in living things and the patterns of correlation formed among them over time might result in knowledge of a distinctive type, but that knowledge reflects nothing about any a priori human expectations about living things. A variety of general learning devices, given the kind of information associatedwith living kinds, would tend to represent that information distinctively and then sequesterthat information into a coherentdomain of biological thought. .Such an approachwas implementedin a connectionist model by McRae, de Sa, and Seidenberg(1997). From a large body of feature norms, they found that. living kinds had a greater density of intercorrelatedfeaturesthan

artifacts and that this distinction could explain the results of severalpriming studiesthat revealedlatency differencesbetween the two kinds. Furthermore, they claimedthat an attrador network that had distributed knowledge of the empirically derived feature norms could simulate the samesort of behavior in priming tasks. Of course, priming results alone are hardly enough reason to supposethat the correspondingliving kind conceptswere fundamentally different in structure, but the work done by McRae and his colleaguesillustrates how suchan argumentmight proceed.! Alternatively , it might be exceedingly difficult to specify any objective informational patterns that set aside plants and animals as a distinct kind with its own rich internal structure; instead, humans may carry with them certain cognitive blasesto interpret information about speciesin highly distinctive and unique ways. Thus, extremely simple perceptualfeaturesshared by animals and plants (perhaps a kind of &actal structure seen more commonly in living things) might trigger a cascadeof predeterminedcognitive blasesthat make learning and the resulting conceptsof biological kinds radically different &om oth~r conceptsof kinds. There may be some real informational differencesas well, suchas differing densitiesof featureclusters, but this view would argue that those differencesfar underdeterminethe nature and kind of specializationsseenin biological thought. We argue that there may be misleading assumptions and misconceptions about the nature of the distinctive information and about the blases that color our notions of what biological thought is like; also, the ensuing misconceptionsabout biological thought, especially as a kind of intuitive theory, may obscure the true nature of speciesconceptsand how they are embeddedwithin the broader system of biological knowledge. Severalnew lines of empiricalwork are suggested. Much of our discussionexamineshow biological thought in general and speciesconcepts in particular develop in the child. We take such a developmenta perspectivebecause(1) it tells us what sorts of information might be most salient to the naive mind, and (2) it suggestswhat might be the most fundamentalblasesthat we all have in thought about the living world. Between the empiricist and nativist extremes there are many gradations and combinations. Our goal here is not to allocate responsibility to these two extremesas much as it is to explore what eachmay contribute to a more complex interactive model of where speciesconceptscome &om and what makesthem special. We do so by trying to clarify what informational patterns might be both distinctive to living kinds and salient to humans, and by asking what sorts of cognitive blasesmight interact with that information, even if those blasesthemselvesare not always reservedexclusively for biological phenomena. THE GENERAL NATURE OF FOLKBIOLOGICAL mOUGHT " Any notions of speciesare embeddedwithin broader systemsof belief about biological kinds. One cannot understandwhat a dog is without also knowing

264

IV. Speciesin Mind and Culture

something about animal life cycles, nutritional needs, and the like. For.that reason, it is helpful to consider how views of intuitive biological thought have changed in recent years (e.g., Medin and Atran 1999). It has become commonplaceto argue that lay people throughout the world possessrichly structured beliefs about the living world that might be thought of as intuitive or naive theories about biological processes and systems- such as , and death ( Medinand Atran 1999). growth , reproduction, digestion, disease beliefs as theories or mental models, however, Characterizationsof these may carry with them somewhatmisleadingideasabout how that knowledge is representedin the mind of the individual. It might seemthat the knowledge must be an explicit set of beliefs connectedtogether in the tightly coherent manner of a formal scientific theory and that the models must contain concrete , imagelike components whose interactions can be clearly visualized. Folk theories would then be said to differ from formal scientific ones only in terms of the particular setsof beliefs they embraceand not so much in terms of their generalformat. Thus, a belief that demonic possessioncausesdisease contagion might have very much the samekind of mechanisticset of lawful relations as a belief that germs causediseasecontagion. (Keil et al. 1999). But a closer look at intuitive biological theories and perhapsat many theories in the more formal sciencesreveals something quite different from an explicit set of propositions all linked together in a tightly connected, logically consistent, and coherent set of inferences. Most people have strikingly little knowledge of the detailed mechanismsat work in their own bodies, let alone in other animals and plants. An exceedingly simple gloss may be all that is known, such as food contains energy and that the body uses that energy as a kind of fuel to power muscles, which makeus move. This simple functional schematamay then occasionallyget filled with local mechanisms and gradually becomeinterconnectedin somewhatlarger and more coherent structures, but only the smallestpercentageof people in any culture can tell you much of anything about the full causalchain that goes from the ingestion of food to the production of a motor movement basedon the energy in that food. . They Yet people seemto have far more than a set of functional schemata classes of of abstract and often seemto have general ways choosing among explanations about biological phenomena, even though these explanations may be equally satisfactoryfrom a functional point of view. An explanation of mechanismsof digestion in "terms of a mechanismin which food particles are converted to light that is routed around the body before being transformed into muscle energy would be vastly less plausible to most adults than a mechanismthat invoked transformation of food into a kind of fluid " fuel " even if both mechanismswere , ultimately wrong . above the level of specificmechanisms of sorts , are several There things, kinds about and believe that adults and possibly children know biological and. that nonethelessmight be distinctive to those kinds and thereby make biological thought different:

265

/ K~il and Richardson: Species of Causation , Stuff, andPatterns

1. We know that certain kinds of properties and relations tend to be central to explanations in biology and to the stability of various biological phenomenad . Boyd, chapter 6 in this volume). For example, (a) color is likely to playa more important causalrole in explaining functions of most living kinds than in explaining functions of most artifacts; (b) size variation might matter less for most living kinds than for most artifacts; (c) difficulty of Bnding instancesmay be more important for artifacts; (d) sensitivity to temperature is more important to living kinds, as is (e) th~ir age or stage of development. 2. We know that certain kinds of causalpatternings might be distinctive to living kinds and to explanationsabout their nature. Patternsof causalhomeostasismay appearto be richer and more interconnected(Boyd, chapter 6 in this volume). Causalmechanismsmay changemore dramatically as one goes from " inside" a biological kind to its outside, but not for artifacts and nonliving natural kinds. We may expect the time course of bounded causal events for biological systemsto have a certain duration that is distinctive in comparisonto the time .coursefor either artifacts or nonliving natural kinds. Artifacts tend to have immediatecauseand effects, whereasliving things can have a far more delayed reaction to events; for example, a plant may not ' yield fruit now becauseit wasnt watered enough a month ago. 3. We know that living kinds may have certain causalpowers even if we don' t know the mechanismsfor thesepowers or their functions. For example, we might know that humanstend to sneezewhen they go rapidly from dim light to bright light , but we may have absolutely no idea how that happens or for what possible reason. The specmccausalpowers of living things may be quite different in nature and kind from the causalpowers of other sorts of things and could thereby give a distinctive structure of biological thought without knowledge of mechanisms . 4. We may also carry distinctive blasesabout aspectsof biological entities and events- blasesthat may not be correct, but that powerfully constrain our beliefs and explanationsabout biological phenomenanonetheless.These blasesmight include the notions that living kinds have fixed inner essences that guide the expression and maintenanceof many .of their phenomenal . and that their properties properties are likely to be present for functional " " reasons . The essentialist bias for living kinds becomesparticularly adaptive important later when we examine the developmental course of naive folk biology . s. We may have notions of how aspectsof biological knowledge might be distributed in the minds of others such that we believe that there are people who know certain things about living kinds and thus can answer our questions . Becausewe cannot understandall the details of most mechanisms , we all learn to divvy up knowledge responsibilities, and part of our understanding of a phenomenon becomesknowing how to accessrelevant areas of expertise that others have generated. Biological knowledge and most other

266

/ IV. Speciesin Mind and Culture

areas of scientific understanding have this critical social component ( Wilson and Kei I1998 ). Notice , however , that historically and in most cultures even today , the richness of and social apportionment of knowledge about the biological world far exceeds the richness of knowledge about the rest of the natural world , such as elements and compounds or weather systems or the stars. The cognitive division of knowledge responsibilities may therefore be far and away the most developed in the biological realm. It is possible that we may have different expectations about how such knowledge is distributed for biological kinds than we do about how it is distributed for other sorts of kinds . For example , it may be that we expect natural kind expertise to be more tied to species or types and artifact expertise to be related to mechanisms or processes. Thus , we have vets , gardeners, pediatricians , and entomologists , on the one hand , and electricians , plumbers , and carpenters, on the other . It seems plausible that this distinction is because knowledge of living kinds has historically begun with the organism and analyzed inward , whereas expertise about nonliving things would consist of grasping a process or mech~ sm and learning to exploit it for functional means across types . Moreover , as science progress es and we learn more about biological mechanisms, we can analyze them more in terms of pro " " cesses. This distinction may therefore be dissolving as cross-type domains of living kind knowledge develop , such as microbiology or evolutionary biology . There are many forms of implicit knowledge one could have of living things . Explanatory knowledge of most phenomena can, by necessity , capture only part of the richness and complexity of causal structure in a domain . We therefore cannot ever have full mechanistic knowledge , but at the same time we have more than a collection of surface impressions or skeletal functional schemata. That our knowledge and understanding might be largely organized in these different forms , however , is not always appreciated , and part of the reason it is difficult to see may be that we all tend to have a vividness illusion regarding our own understandings . People may often assume they have complete or extensive mechanistic understanding of a domain when they do not . They might have observed the inner workings of a car engine or a heart or a bicycle derail leur and be convinced that they have an imagelike mental model of how that system works . They might confidently predict their ability to explain exactly what happens in each step of the causal event sequence that characterizes the entity in question ; yet when querled about such mechanisms, they might reveal glaring ignorance and inconsistencies . They seem to think that understanding arises from clockwork kinds of vivid concrete steps, and because they have a strong sense of understanding , they assume they have that kind of clockwork knowledge . Later , we discuss how the vividness illusion may help us understand why mowledge does not always develop from concrete images of interacting components to more abstract notions about property types and their interactions .

267

/ : Species Keil and RiChardson , Stuff, and Patternsof

Just as we tend to overestimate the centrality of such clockwork mechanisms in our adult understandings , we may tend to overestimate their seminal role in development . In the end, any notions of species must be powerfully influenced by how we view adult biological knowledge in general . If , for example , we believe that having an essence is a critical part of a species concept , then we must be able to say how we could have a notion of essence without any sort of concrete understanding of what is inside animals and plants . And if we think species are entities with properties that help maintain the survival and integrity of that kind , we must be able to say how we know those patterns of maintenance without knowing many of the details . Almost any way one tries to flesh out the nature of species concepts is powerfully influenced by this view of biological knowledge in general . The species concepts cannot be understood without understanding how it fits into a larger system of folk biology .

CHILDR~ 'I, FOLK , ANDSCIENTISTS ' Lay people are not children relative to scientists; thus , the study of children s biological concepts is not equivalent to studying biological concepts in ' people who don t work in the science of biology . All adults have many interactions with the biological world . Even the most nature phobic and jaded urban dweller thinks about the food and drink he consumes and their consequences on his bodily functioning , or about the diseases he encounters in others and how they might influence his own biological state. Adults have spent thousands of more hours having such thoughts than a young child has, and they have surely developed much richer and more elaborate systems of knowledge . As we look across many different cultures , however , we turn to development as highlighting emerging and invariant universal properties of biological thought , for those properties that seem to emerge the earliest might well be the most universal . They may have much more elaborated forms in adults , but examining acquisition may help uncover a core form . Views of the emergence of biological thought have changed dramatically in recent years . Older accounts described children as going through general stages of cognitive development that would make completely unavailable to them any real notions of biology . Piaget ( 1954 ), for example , thought of " " young children as having animistic tendencies in which they endowed a great many things , living and not , with beliefs and desires and explained their properties and actions in such terms. They did not have any notion of plants and animals as forming a common kind of biological things . But these stage views of cognitive development fell from grace in the late 1970s and early 1980s as closer and more systematic examinations of knowledge acquisition did not reveal such stages in which children progressed from having one sort of representational and computational capacity to a qualita -

268

IV. Species in Mind andCulture

tively different one (e.g ., Gelmanand Baillargeon 1983). Alternative conceptions focusedon developmentalpatterns in specificbounded domains- such as biological thought, theory of mind, naive physical mechanics , and number. This domain specificity approachfound a natural affinity with the notion of conceptsas embeddedin larger systemsof explanation or intuitive theories. " " Despite occasionaltalk of theories of everything, the notion of theory tends to imply a bounded domain of phenomenaexplained by that theory with little reasonto think that theoretical knowledge in one domain should necessarilyextend to all other domains. Within that framework, two main themesemergedwith respectto the emergenceof biological thought. One theme argued that younger children (e.g ., before age seven) really did not have an appreciationof biology as a coherentset of phenomenaand that they often explainedbiological phenomenain socialand psychological terms (Carey 1985). Thus, they might see sleeping as causedby feeling tired and wanting to sleep, and they might see its function as satisfying those needs. Similarly, they might explain eating in terms of the sensory pleasuresof ingesting food and in terms of the social interactions that happen at meal times. They would completely ignore the physiological aspects of these processes. In these accounts, the children would then undergo radical conceptual changesin which an intuitive biology emerged, and now they would explain the samephenomenain completely different terms. An alternative accounthas emergedfrom severallaboratories(KeiI1992 , 1998; Hatano and Inagaki 1996; Wellman and Gelman 1998), however, that suggeststhat even young preschoolersdo have some senseof the domain of biology and have a distinctive mode of biological thought. Their failures on many tasks concerning biological knowledge are attributed to their not knowing specific mechanismsand to sometimesmisunderstandingwhich frame of reference (social versusbiological) is being askedfor in a task (Gutheil, Vera, and Keil 1998) rather than to a complete lack of an intuitive biology , which is seenas couchedin the more implicit forms discussedearlier. In addition to preferring some mechanismsmore than others, young children also often know how knowledge of a specificmechanismis likely to be related to understanding of another mechanism , even when they have no direct knowledge of either mechanism . For example, in one study in progress in our laboratory, children as young as five years of age often respond in an adultlike mannerto the following question: Louise knows all about why kids get a secondset of teeth. Cathy knows all about why babiescan get afraid of strangers. Who knows why teenagerslike to listen to so much music? Many five-year-old children judge that knowledge about behavioral " mechanismsis more likely . to hang together" in the minds of experts (e.g ., Cathy is more likely to know about the teenagerspredilection for music) than will a mixture of biological and behavioral mechanisms(e.g., where Louise would know more about the teenagers) even though they usually

269

/: Keil and RiChardson : Species , Stuff, and Patternsof Causation

have no explicit notions of the mechanismswhatsoever. Indeed, when asked about the mechanisms , most of them will quickly say they have no idea. Yet, at somelevel thesechildren are aware that the kinds of processes involved in stranger fear and positive feelings toward music are more similar to each other conceptually than those processes involved in explaining deciduous teeth. The conversealso seemsto hold- namely, if there are two questions about biological mechanismsand one about behavioral mechanisms , an expert in one of the biological questionsis judged more likely to know more about the other biological question than about the behavioral one. These sorts of results raise the question, What do young children know about biology that guides their judgments in suchtasks? More generally, it seemsthat we do not want to think of young children as only capable of having " concrete" impressions of biological processes wherein specificmechanismsor models are explicitly visualized. Contrary to many decadesof claimsthat younger children are bound to think in concrete terms and can grasp more abstract relations only later, it often seemsthat knowledge can shift from the abstract to the concrete in development (Simonsand Keil1995 ). We adopt here the view that even preschoolershave an intuitive sense, often at a highly implicit level, of biological phenomenaas being distinct from one another, and we ask what that perspectivesuggests about speciesconceptsand how they develop. One of the most robust findings of that developmentalstory hasbeen the discovery of very early beliefs in essencesfor living kinds. How might a child come to believe more strongly in essencesfor living kinds than for other sorts of things? We've seen that such a strong belief might arise not only from real informational differencesbetween the causal structures responsiblefor biological kinds and those structures responsible for other sorts of kinds, but also from cognitive blasesthat might be related to those differences. Yet another sourceof information might be parents, but in a surprisingly subtle and implicit manner. Parentsdo not tell children that , but they do talk about living kinds quite differently things have essences from the way they talk about artifacts and in ways that would seemto suggest a hidden structure with greater richnessfor living kinds (Gelman et al. 1998). Parentsdo not provide didactic explanationsof hidden properties and their causalconsequences . Instead, they seem to indicate in more abstract different ways through patterns of reference- what sorts of things are kinds and that some kinds are likely to have richer essences . They also provide hints as to what sorts of things are more likely to be taxonomically embedded. . We do not yet' know how much influence this subtle pattern of language has on the child s conceptual development, but it does point out a third dimension that might interact with cognitive blases and intrinsic informational differences. We now need to examinethe content of this essenceconcept that adults and children readily attach to living kinds.

270

/ Speciesin Mind andCulture

THEFSSEN CEOFFSSEN CE All of us may succumbto essentialistblasesthat compel us to assumeand look for those critical aspectsof kinds that allow us to make powerful inferences . Gelman and Hirschfeld (1999) contrast three types of " psychological " : sortal ideal and causal. Sorial essence essentialism refers to critical defining , , featuresor, in other words, singly necessaryand jointly sufficient setsof features for determining category membership- an accountthat seemsto work for only a small set of real-world concepts, however. Ideal essence refers to nonexistent perfect cases. The ideal essenceof parallel lines has no real counterpartsbecauseno physical system can perfectly embody parallelism. refers to something about a kind that results in its Finally, causalessence of its most having many typical and stable properties. The nature of that " " is critical. Most commonly, it seemsto be thought of as a fixed something inner entity that has multiple causaleffects. That entity might be a kind of substance(the essenceof gold being atoms with gold' s number of protons) or an informational code (the DNA sequences] correspondingto tigers) or a process. Despite much talk about essencein recent years, it still is not clear how most lay people actually conceive of essencesor whether the blases are much stronger in younger children and in thought about somesorts of kinds rather than about others. Medin and Ortony (1988) have suggestedthat we can often believe in an essencewithout any idea of what the essenceactually " is- that we have an essence place holder" concept without the concept of the essenceitself. This senseof essencemakesclearerhow it could be a bias without specificcontent. But the place holder notion may overlook a related set of beliefs, for even as people commonly believe in essenceswithout , they still might prefer some sorts knowing any details about those essences of future details to others. Thus, there might be a physicalist bias, whereby " " people prefer essencesto be seenas objects or stuff rather than kinds of processes. Teaching that photosynthesisis the essenceof green plants might be less compelling than teaching that it is a certain DNA sequence , even though the processof photosynthesismay be much more directly connected causallyto a far greater range of phenomenalproperties of plants.1For speciesconcepts, then, notions of essencemay be absolutely essential, yet we have little idea of what sorts of constraints there might be on . Sometheorists have argued that notions of essencehave notions of essence had extraordinary limiting effectson how we think about speciesin the context of evolutionary thought. Hull (1965), for example, talks about how " " Ari .stotelianessentialismcauseda 2,000 year stasisin evolutionary thought becauseit assumedthat specieshad fixed essencesand thereby could not explain how new speciescould evolve through natural selection. The notion of speciesas a probabilistic concept, a distribution of types, seemedto be foreclosedby the essentialistbias even though that notion of distribution is

271

/ Keil andRichard son: Species , Stuff, andPatternsof Causation

critical to understanding how evolution through natural selection could actually occur. But what notions of essencewere involved in Aristotelian essences ? Hull ' s reference to Aristotle suggests something like the sortal essence , yet for much of folk biology , the causalessenceseemsmore appropriate . A causal essence , however, is not on its own incompatible with a notion. Indeed, Boyd' s notion of causalhomeostasis(seechapter probability 6 in this volume), a process wherein speciesproperties are maintained in stableconAgurations, fully allows for speciesthemselvesto changeover time as a consequenceof natural selection, yet that kind of processmight well be a form of causalessence . The fixednessof essencewould seentto arise from a cognitive bias toward not appreciating a process such as causalhomeostasis, either becauseprocesses in general are not preferred or becauseany probabilistic componentsto suchprocesses are not allowed. The problem of an essenceis that it seemsindirectly or probabilistically related to the features of an organism , but directly related to its categorization " will . That is to say, an organism with a leopard " essence resemble a leopard ceterisparibus, but not if some freak dietary anomaly prevents it from developing a certain pigmentation pattern: it will neverthelessbe a leopard under an essentialistconception. The question, then, is what features licensethis ascription of an essenceof a certain type, sincean essenceis also a basisfor ignoring features? The riddle about causal essenceis that unlike both sortal and ideal essences , causalrelations in the real world are rarely strictly necessary . Because the entities within a folk theory are usually loosely framed, there can be no necessarycausallaws holding between the entities as there can be between entities in a more rigid scientific framework. Therefore, some degree of probability always seemsto be associatedwith causalrelations occurring in the real world , even if in some casesthat probability is exceedingly high. Are lay notions of causality ignorant of suchprobabilities, or does probability somehow otherwise get ignored as notions of causeand essencebecome intertwined? Peopleoften seemto know that an event may causea particular effect, but not always. Eating rotten meat usually causesone to feel sick, but not always. Sexual relations causethe emergenceof babies sometime later, but certainly not alw,ays. These probabilistic relations are part of how all of us talk every day: why , then, cannot there be a kind of causalessencethat is probabilistic? It would seem that the psychological constraints on causal essencestrongly discourage certain forms. Perhaps the psychologically appealingsenseof causalessenceultimately requiresthe notion of fixed stuff, rather than a process, as the initial cause, and even if that stuff has only , its very nature is not at all probabilistic. . probabilistic causalconsequences Given these considerations, there is the additional problem of the scope of an essence : how 'it is seen to vary within living kind hierarchies. Is the " " essence a type- the DNA sequencevarying by " product type and brand" like a supermarketbar code - or is it a token, an essenceas individual as a fingerprint? Or is it the case that the hypothesized essentialistbias easily

272

/ IV. Speciesin Mind and Culture

admits gradations of similarity between tokens that can be used to categorize them into types? Although it may be more correct to think of a continuum of DNA similarity that is, perhaps arbitrarily, divided up into , speciescognitive bias may pull us away from this idea. In their pioneering studies on prototype concepts, Rosch and Mervis 1975 ( ) asked subjects to grade the typicality of various living kinds. For example, a robin was graded as a more typical bird than a duck, and these ratings predicted speedon tasks such as lexical decision. Furthermore, Rips (1975) found that the typicality gave rise to asymmetricjudgments: subjects thought it more likely that a duck could catch a diseasefrom a robin than that a robin could catch a diseasefrom a duck. This asymmetry could be interpreted as pointing toward an essentialistbias. Rather than it being the case that robins and ducks simply have a certain degree of similarity, it seemsto be the casethat we conceive of robins as having a " stronger bird " essence than ducks, so if something affectsa robi,nas a typical bird, it seems more likely to affect a duck than vice versa. An essentialistfolk theory may suggestthat biological factors suchas susceptibly to a certain diseaseattach not only to robins, but to their essenceas birds; thesefactors are hence, more likely to extend to a duck. Can the same senseof essencethat gives us the concept of speciesalso make one species more central than another in a higher-level category, suchas bird? These issuesmake obvious the need for an extensive set of psychological studies that ask what constraints there might be on different notions of essence . We can further askhow those notions might changeover the course of cognitive development in the child- of moving from novice to expert knowledge - and how they vary acrosskinds. In addition, they may vary acrosskinds from the earliestpoints in conceptualdevelopment, or they may start as a more common vague notion that gradually differentiates with increasingknowledge in each domain. In short, people do seemto have an essentialistbias and perhapsepecially so for living kinds, a bias that power, but we have only begun to understand fully influencestheir conceptsof species the real psychologicalnature of this bias.

THE VIVID ILLUSIONOF SPECIFS So far we have put forward several claims about the nature of our concepts of living kinds , and it seems that there might be a common underlying cognitive explanation . Given that folk biological thought seems both void of specific mechanisms and inclined toward certain types of explanation , and that we have a strong bias toward essences in living kinds , it now seems plausible to seek an account of the type of illusion discussed earlier , whereby people have a tendency .to assume that they have a vivid , clockwork knowl ~ ge of certain mechanisms. This illusion is similar to one that has been repeatedly demonstrated in recent ye~ s in studies of visual memory and indeed may arise from common

273

Keil and .

dson: Species , Stuff, and Patternsof Causation

mental sources. In those studies, people look at scenes , often quite simple ones, and assert that they have a clear memory of what it contained. Yet when tested, they can be strikingly ignorant of the details of the scenethey just observed. They not only mistakenly rememberdifferent colors, textures, and surfacepatterns on objects, but also often fail to notice in a recognition task when completely different objects are in the scene(e.g ., Simons 1996). They do, however, have quite good memoriesfor the spatial layout of the objects in the scene- the general relational topology of objects- even when they forget the details. With dynamic objects and systems, people seem to retain good understandingsof the functional 'layout " while often losing all the details of particular discretecomponentsin that layout . The vividness illusion in the mechanismsof folk biology might be seen in terms of a misleading dispositional bias: a variant of the " fundamental attribution error" ( Nisbett and Ross 1980) whereby folk tend to think dispositionall about other people, assumingthat inner essences , more than situational factors, explain the behaviors of others (Miller 1996). This bias has been shown to extend beyond conceptsof people to conceptsof other entities in the world , such as a chip of wood in a turbulent stream (Peng and Nisbett forthcoming). It may be that living kinds are far more powerful triggers of the bias than are most other kinds. A very general cognitive bias may be at work here as well: the tendency to focus on what are known in statistics as main effects and not on interactions . It may be simpler and more cognitively compelling to think of a kind being created by either intrinsic essentialproperties or environmental forces, rather than by an interaction between the two . Therefore, being aware of some salient endogenous factors may lead to an overzealous assumptionof almost exclusively endogenousforces. The reasonsfor these illusions or blasesare not clear, but they may well have cognitive benefitsat somelevel. For scenes , they perhapshelp build the of a continuous flow of impression experience; for systems, they may help build an impressionof a continuous chain of understandingwithout explanatory gaps. The question with regard to speciesand essentialismis whether a kind of illusion is created wherein patterns of causalhomeostasisresult in the relative stability of property clusters, which are then mistakenly assumedto be stable not becauseof that homeostaticprocess, but rather becauseof a fixed physical causal source. That is, causal homeostasiscausesstable property clusters, which in turn causethe impressionof a fixed physical essence . We know that people succumbto a vividness illusion in severalways. There may be the corresponding assumption that stable property clusters must have . stable physical sources. The relative stability of property clusters that , as opposed to those clusters that do emerge through causalhomeostasis not, may be so great in relative terms that it leadsto the erroneousassumption of absolute stability . Then, the cognitive bias toward essencesmight

274

N . Speciesin Mind and Culture

consist of positing a stable property for living kinds as a kind of simplifying heuristic. - that Many argue that living kinds have a much richer causalstructure That causal . al. et forthcoming) they are causally more complex (Gelman . essence of causal complexity is then thought perhapsto trigger impressions But complexity is notoriously difficult to define and measure, and a little reflection makesone worry about any absolutedifferencesbetween artifacts and living kinds in terms of causalcomplexity. To be sure, the vast majority of living kinds have more complex causalinternal processes that give rise to surfaceprocesses and to activities. When one includes external social and cultural factors that help explain why artifacts are as they are, however, artifacts have vastly more causalcomplexity connectedto such cultural and social properties than most living kinds have. be at work. The Again, more subtle and more interesting differencesmay socioculturalcausalfactors for artifacts may not be nearly as bounded as are the other causalfactors for most living kinds. That is, they do not neatly circumscribethe artifact. To know why chairs are the way they are, one has to look at economics, body shapes, and physiological needs in a vastly extendedcausalnetwork that does not cluster tightly around chairs. Whereasartifacts may have causalfactors that are distributed acrosssoci ocultural factors, living kinds seemto have a more bounded, visceral pattern of causalhomeostasis. Perhapsliving kinds are different in the respect that the causalcluster for each one is more of a denseisland in a sea of weaker and lesscausallycomplex interconnections. Again, there is a strong intuition here, but it needsto be examinedin an experimentalmanner to seeif people see living kinds as forming more dense and distinct clusters (Seealso Ahn , 1998 on why different featuresare central for artifacts and natural kinds.). We all accept that the notion of essenceis not unique to thought about for living kinds even living kinds, but its strength and power seemstrongest in such cases. Thus, at fixed of entity when it may be least correct as a kind the cognitive level and perhapsalso becauseof the specialnature of living character, one that things, essencesof living kinds seemto have a different may be more cognitively compelling. , but again that Artifacts are not normally thought of as having essences notion dependson notions of essencethat remainlargely unanalyzed. Certain devicesphysically physical constraintsmakeonly sometwo wheeledpedaled . The angle of the 1990 and Olson ) as useful ( Kyle stableand thereby bicycles " are " the that completely unridable as bicycles &ont fork can be changedso a Thus , pattern of causalhomeothey start to oscillatein an unstablemanner. stasismakesa stablefunctional unit, and only certain properties qualify in the ? If of the essenceof bicycle pa'ttern for that unit. Are suchcausalpatterns part not , what makestheir caseso different &om the patterns for some animals? lie in the belief that for living ~ t a psychological level, the difference may kinds, there is some sort of fixed stuff that gives rise to the patterns of homeostasis , a belief that seemscounterintuitive for most artifacts.

275

, Stuff, and Patternsof Causation Keil and RI 'chardson: Species

SPECIES CONCEPTS AS A DISTINCTKIND OF CATEGORIZADON A speciesconceptis a kind of categorization . It treatsa classof living things asequivalentin importantrespects , andthat equivalence thenlicensespowerful inductions , whichis presumably why species conceptsaresouseful.But inductivepower is said to be a key motivationfor almostall cognitively naturalcategories ; what else, beyondessence , is distinctiveabout species ? Thereappearto be manyqualitative conceptsasopposedto otherconcepts distinctionsbetweenliving kinds and artifacts , includinghow taxonomies , teleology, and exemplarsare construed , yet the questionremainswhether or not thesedifferences aredue to quantitativedifferences in the spreadof causalhomeostatic . patterns Living kinds are saidto be muchmore deeplyembeddedin taxonomies thanareothersortsof kinds, andthroughoutthe world, all peoplesseemto realizethis taxonomiccharacter on in very early (Atran, chapter development 9 in this volumeand 1998). Abundantevidencenow showsthat this taxonomicassumption is very powerfulfor living kindsandprobablynot nearly aspowerfulfor othersortsof kinds(althoughthe latterhavenot been nearly ascarefullyor systematically studied). However, whatis it aboutliving kinds that leadsto this taxonomicassumption ? As with essences , it is not so clear what "triggers" the assumption . Thus, it is not so objectivelyobviousthat living kindsoccurin moredeeplyembedded hierarchies . Thereare, afterall, quite deephierarchiesfor many naturallyoccurringcompounds(kinds of rocks, soils, gems, etc.). For example , the UnitedStatesDepartmentof Agriculture a four basic gives layered taxonomyof soil types- categorizing , to takeone instance from histolsolsto fibriststo sphagnofibrists , that areper, depth of hierarchyis not evidentlya cue. Similarly, there gelic. Therefore arevery rich hierarchies for manyclasses of artifacts.The U.S. PatentOffice has more than five hundred classesof artifacts : meanderingdown the " has scheme , of thesefive hundred , the class"Surgery forty divisions,within " has which "Prosthetics within which "Leg" hasnine, and of these , twenty "Socket " hassix . Evenif the hierarchies of living kindsaregenerally categories , it is not at all clearthat their depth in itself is responsiblefor deeper their being seentaxonomically . Thus, thereis no evidenceto supportthe intuition that deeperhierarchies aremoretaxonomicallycompelling . If depth doesn't predicta senseof taxonomyfor the nonlivingworld, why shouldit be a factorin theliving world? -relatedtaxonomies Do thesespecies provide the only apparentways to kinds eventhoughtherearemultiplewaysfor other sortsof classifyliving kinds? No. Animalscanbe classified aspredatorsor prey, asdomesticated or wild, or asedibleor not, or they canbe classifiedwithin theoriesthat focus on phylogenetics , ecology, and so on. It is difficult to know how to count . the ways, but it is not obviousthat . living kindsaredifferentin this respect . Evenin folk biology, we seetreesand nontrees asone way of sortingplants

276

N . Sp~

s in Mind and Culture

that is at intrinsicoddswith sortingby morecommonspeciesnames(Dupre 1981). Do taxonomiesfor living kindshavemoresalientpropertiesat eachlevel to nonlivingnatural ? Perhapsthat is the casein comparison of the hierarchy featuresat many distinctive of kinds, but manyartifactshavepowerfulsets . Perhapsthere is a levels, and those featureslicensepowerful inductions -levelconcepts(Roschet ale1976) turn out moresubtledifferencehere. Basic - suchasfurniture, clothing, and artifacts some for to be muchmorerobust . That is, the basiclevel for kinds vehicles than they do for most living a vastlyricherfeatureclusterthanis presentat artifacts(e.g., chair) represents level(e.g., furniture); the basiclevelalsoofferscorrespondingly the superordinate . With living kinds, however, the superordinate manymoreinductions level, suchasmammal , is muchricherin natureand morefull of distinctive more propertiesthanit is for mostartifactkinds. Thisdifferencemaysuggest of less thereis taxonomiccontinuitybetweenlevelsof living thingsbecause , a jump in featuredensityasonemovesup or down a level. For that reason artifact for relief in much out categories the basiclevel tendsto stand sharper "data that level morestronglyand demarcating , with most experimental whenartifactsareusedasstimuli(Roschet ale1976). unambiguously for otherkinds andcategories Anothercontrastbetweenspecies categories that the teleological argumentbeing is said to revolve aroundteleology, however form , this a In . , work only with living kinds simple interpretations . their , as do parts In a argumentis clearlywrong. Artifactshavepurposes moresubtleform, however, this view may offer an importantcontrast.One doesnot normallyaskaboutthe purposesof animalsaswhole entities, but domesticated only aboutthe purposesof aspectsof them(the exceptionis highly are functionalspecies , suchas hunting dogs). Suchquestions completely aredifferent categories , however.Thus, species normalaboutartifacts when and directed are they are in terms of where teleologicalquestions unclear remain is , it universal how and exists legitimate.Why this difference however. andsomecharacterization of teleologicalunderstandings Detailedanalyses holistic such blocks that kinds teleologicalinterpretations is about it of what living kindsto and between contrast the nonliving areneededfor living the intention human , guides holistic work (d . Bloom 1996). With artifacts , but why shouldintention be neededfor the holistic teleologicalaccount accountwhen it is not neededfor more local explanationsfor living kinds , that webbedfeethelpananimalswimbetter)? (for example A Analcontrastmay circlearoundthe natureof the categorystructurein . With somenaturalkinds, the termsof bestexemplarsandwell deAnedness ' , to useGelmans best exemplarwould be somethinglike its sortalessence " " . The , for example terminology: pure water is nothingbut H2Omolecules in framed essence ideal some to closer be bestexemplarof an artifactwould fast be , would , easy cheap termsof its function, so the exemplarof computer , and so on. However, determiningthe to use, compatiblewith all platforms

277

: Species,Stuff, and Patternsof Causation Keil and chardson

bestexemplarbecomes moredifficultwhenthe functionis not easilyhamed: What is the one-line functionaldefinitionfor churchor theinternet ? How do we think of the exemplarsof living kinds? In somecases , living kindshavefunctionalvaluefor humanbeingsand canbe evaluatedon that dimension : the besttreesarethosewith the highestfruit yield, for . example " : the Or we might hamethe bestexemplarasthat whichfulfills its " purpose bestbutterfly is camouflaged Hornpredators , can extractnectarefficiently , lay a largenumberof eggs, and so on. But this hameworkbringsin all the problemsof teleologydiscussedabove. In addition, how do we evaluate how "doglike" a dog is? It couldbe the casethat we havesome prototypeor idealizedaverageof all dogs. Doesthis averageequatewith a ? dog essence Alternatively, we may havesomenotion that goesbeyondthe typicality of of a homeostaticclusterof causal perceptualfeaturesto a representation . It becomesdifficult, however, to propertiesthat are seento be significant drawa firm distinctionbetweenliving andnonlivingkindson this criteriaof well definedness because thereis considerable variationwithin both classes . Well definedness in itself may thereforenot be the cut betweenartifacts andliving kindsbecause they both havefuzzyaspects , but the fuzzlness may occurin differentwaysfor the two kinds. Furthermore , it seemsthat when morerigid criteriaareappliedin the evaluationof living kinds, thosekinds arebeingmeasured in relationto somefunctionalgoal, eventhe seemingly hollow purposeof winning a dog show. In short, species conceptsmay be importantlydifferentHornother sortsof conceptsthat categorizekinds, but the dimensions anddegreesof differencearestill unclear . It seems , therefore , that speciesconceptsmay be distinctin severalways that go beyondnotionsof essence andthat arenot closelyrelatedto patterns of causalhomeostasis . The detailsof the psychologicaldifferences between thesewaysarejust beginningto be uncovered , however.

CONCLUSIONS Speciesconceptsseemindeed to reflect a specialkind of categorization. Several cognitive blasesmay interact with some very distinctive informational patterns of living kinds, such as a disposition toward certain types of properties and relations, as well as a tendency to discretize these homeostatic patterns into an essenceof some sort. But it is clear that all of thesepsychological contrasts are just beginning to become apparent. The nature of a speciesconceptis mostly a placeholder at presentand is &amedby only the softest and vaguest of constraints. We therefore do not really know much about how it is that speciesconceptsarise, but this lack of knowledge should . not be discouraging becausefor the first time we are now in a position to learn a great deal more about how those concepts emerge in development and become used by adults. The rapid growth of work on biological thought, especially in cross-cultural and developmental perspectives, has helpe.d set up a &amework in which it is now possible to pose highly

278

N . S~ esin Mind andCulture

detailed questionsabout speciesconcepts- such as how children have them early on, even though they have few concrete details of what an essence might be. We now seemore clearly the different ways in which folk biology is likely to be mentally representedand especiallyhow it is not. Even more important, we now seehow these issuescan be better understood through of results empirical study of both children and adults. Very different patterns are possible in such studies, and those patterns will make a profound difference in our understanding of what it means psychologically to have the conceptof a species.

ACKNOWLEDGMENTS therein of partsof thispaperandsomeof thestudiesdescribed Preparation to HD23922 ROI Health of grant weresupported by anNationalInstitutes FrankKeil. NOTES ' ) 1. As an aside , it is interestingto note the similaritiesbetweenMcRaeandcolleagues(1997 concepts representing of anattradorspaceasa way of describinga cognitivemechanism discussion ' s (1996) claimthattheconceptof spedesshouldbeunderstood andGoodwinandWebster " seeGriffiths, " Geld ( chapter in biologytheoryin termsof an attradorspaceor a morphonogenic . 8 in thisvolume) 2. Somesort of cognitivebiasmaybe favoringthenotionof DNA in lay thinkingaboutliving in this volume, of living kinds. As arguedelsewhere kindsandappropriatingit as the essence more be can there because variability DNA is not a sufficienttool with whichto dividespecies , it appearsasif the scientifictermDNA hasbeen within speciesthanbetweenthem. However andDNA divideup , eventhoughessence equatedwith someintuitive, folk notion of essence groundsfor this bias, the world in very differentways. We discussthe possiblepsychological the notion of but for now we might makethe observationthat whenpeoplemisunderstand an belie make mistakes underlying the do commonly few a all but they DNA asprobably belief. essentialist

REFERENCES ? Cognition , 69, Ahn, W. (1998). Why aredifferentfeaturescentralfor naturalkindsandartifacts 135 178. andcultural : Cognitiveuniversals Atran, S. (1998). Folkbiologyandtheanthropologyof science - 611. 547 21 Sciences Brain and , . Behaoioral particulars 60, 1- 29. . Cognition , history, andartifactconcepts Bloom , P. (1996). Intention . . Cambridge , MUs.: MIT Press in childhood change Carey,S. (1985). Conceptual 90, 66- 90. Reuiew Dupre,J. (1981). Biologicaltaxaasnaturalkinds. Philosophical . In J. H. Flavelland (1983). A reviewof somePiagetianconcepts Gelman . R, andR Baillargeon . : York New 3. vol. Wiley , , eds., Handbook of childpsychology ~: M. Markman folk and biology. In D. MedinandS. Atran, Gelman , S. A , andL Hirschfeld(1999). Essentialism . Press . : MIT Mass , edS., Folkbi C ?logy. Cambridge

279

, StuH, and Patternsof Causation Keil and F: c( hardson: Species

: GeIman . S. A., J. D. Coley, K. S. Rosengren , andA Pappas (1998). Beyondlabeling , E. Hartman The role of maternalinput in the acquisitionof richly structuredcategories . SRCDMo~ ph Series 253, 1- 114. : Generative andrelational Goodwin . B. C., andG. Webster(1996). FonnandTrRnsformation principles in biology . Cambridge : Cambridge . UniversityPress Gutheil, G., A Vera, andF. C. Keil (1998). Houseflies don't "think" : Patternsof inductionand . Cognition 66, 33 49. biologicalbeliefsin development Hatano , G., andK. Inagaki(1996). Cognitiveandculturalfactorsin the acquisitionof intuitive andhumandevelopment : , eds., Handbook of education biology. In DR . Olsonand N. Torrance NewmtJelsof learning . Cambridge : Blackwe U. , teaching , andschooling on taxonomy : 2,000 yearsof stasis . BritishJournal HulLD. (1965). Theeffectof essentialism for thePhilosophy 15, 314- 326, and16, 1- 18. of Science KeiLF. C. (1992). The originsof an autonomous , biology. In MR . Gunnarand M. Maratsos eds., Modularityandconstraints in language andcognition : Minnesota onchildpsychology , symposium vol. 25. Hillsdale . , N.J.: Erlbaum andtheoriginsof thoughtandknowledge . In R. M. Lerner KeiLF. C. (1998). CognitiveScience , ed., Theoretical models , vol. 1, Handbook , 5th ed. New of humandevelopment of ChildPsychology York: Wiley. andexplanation in the Keil, F. C., D. T. Levin , B. A Richman , andG. Gutheil(1999). Mechanism of biologicalthought: The caseof disease . In D. Medin andS. Atran, eds., Folkdevelopment . Cambridge . , Mass.: MIT Press biology

McRae , K. , de SaV. R., andMS . Seidenberg (1997). On thenatureandscopeof featuralrepresentation of wordmeaning . Journal General , 126, 99- 130. of &ptrimentalPsychology essentialism . In S. VosniadouandA. Ortony, Medin, D. t ., andA Ortony (1988). Psychological . NewYork: Cambridge eds., Similarityandanalogical . reasoning UniversityPress . Cambridge . Medin , D., andS. Atran, eds. (1999). Folkbiology , Mass.: MIT Press Miller, J. G. (1996). Cultureas a sourceof orderin sodal motivation : Comment . Psychological Inquiry7, 240- 243. : Strategies Nisbett, R., and L Ross(1980). Humaninference andshortcomings . of socialjudgment -Hall. Cliffs, N.J.: Prentice Englewood 31, 134. Olson, J. N., andC. R. Kyle(1990). Bicyclestability. Bicycling -culturalsimilarityanddifference inunderstanding ). Cross Peng , K. , andRE . Nisbett(forthcoming . In M. Shale . FrankfortKy.: KentuckyStateUniversity , ed., CultureandScience physicalcausality Press . . , J. (1954). Theconstruction of realityin thechild. NewYork: BasicBooks Piaget . Journalof VerbalLearningand Rips, L (1975). Inductivejudgmentsaboutnaturalcategories VerbalBehavior 14, 665- 681. : Studiesin the internalstructureof Rosch , E., and C. B. Mervis (1975). Familyresemblances . Cognitive 7, 573- 605. Psychology categories -Braem(1976). Basicobjectsin Rosch , E., C. B. Mervis, W. D. Gray, D. Johnson , andP. Boyes naturalcategories . Cognitive 8, 382 439. Psychology Simons fail. Psychological Science , D. J. (1996). In sight, out of mind: Whenobjectrepresentations 7, 301- 305.

280

IV. Speaesin Mind andCulture

of biological Simons , D. J., andF. C. Keil (1995). An abstractto concreteshift in the development " " 56, 129--163. thought:The insides story. Cognition . In Wellman , H. M., andS. A. Gelman(1998). Knowledgeacquisitionin foundationaldomains Child 2 Handbook vol and eds . , , D. KuhnandR. Siegier , perception language Psychology of , , Cognition , 5th ed. NewYork: Wiley. . Minds and Wilson , R. A , and F. C. Keil (1998). The shadowsand shallowsof explanation Machines 8, 137- 159.

281

: Species Keil andI ((chardson , Stuff, andPatternsof Causation

V

Species Begone !

11

Species and the Linnaean

Hierarchy

Marc Ereshefsky

Prior to the eighteenthcentury, biological taxonomy was a chaotic discipline marked by miscommunicationand misunderstanding. Biologists disagreed , how to assign taxa to those categories, on the categoriesof classi6cation taxa name to how even and (Heywood 1985). Fortunately for biology , Linnaeussaw it as his divinely inspired mission to bring order to taxonomy. The system he introduced offered clear and simple rules for construding . It also contained rules of nomenclaturethat greatly enhanced classmcations ' the ability of taxonomists to communicate. Linnaeuss system of classi6ca tion was widely acceptedby the end of the eighteenth century. That acceptance the brought order to a previously disorganized discipline, and it "laid " the the of foundation for unprecedentedflowering of taxonomic research late eighteenth and early nineteenthcenturies(Mayr 1982, 173). In the last two hundred years, the theoretical landscapeof biology has result of changed drastiCally, however. Speciesand other taxa are not the static , timeless divine creation, but the products of evolution. Taxa are not classesof organisms, but evolving and temporary entities. The theoretical of Darwinism assumptionsof the Linnaeansystemhave been replacedby those one these of . In , and the Modem Synthesis light might wonder changes for construding classiif the Linnaean system remains a practical system 6cations. Some authors maintain that it is not (Griffiths 1974, 1976; de Queiroz and Gauthier 1992, 1994; Ereshefsky1994). The processof evolution and the resultant diversity of biota, they argue, render the Linnaean rules of classi6cationand nomenclatureseriously flawed. Somedetractors of the Linnaean system even suggest that it should be replaced (Hull 1966, . Defenders Hennig 1969, Griffiths 1976, de Queiroz and Gauthier 1992, 1994) saving worth is Linnaean the that system of the Linnaeansystem respond 1979 , is necessary ( Wiley , though they agree that some revision and Cracraft 1980, chapter5). Eldredge " The focus of this chapter is not the entire Linnaeanhierarchy, but aparticularly troublesome portion of it : the speciescategory. Two types of problems face the speciescategory: ontological ones and pragmatic ones. For Linnaeus, as well as for the architects of the Modem Synthesis, the divide between the lower and higher ranks of the Linnaeanhierarchy was supposed

to reflect an important ontological divide in nature. As we have discovered, the existenceof that divide is suspect. In addition, Linnaeusand the architects of the Modern Synthesisthought that all speciestaxa are comparablein some important respect. That assumptionis problematic as well. If there is no clear distinction between speciesand higher taxa, and speciestaxa are noncomparableentities, then we have grounds for doubting the existenceof the speciescategory. Theoretical problems grade over to pragmatic ones. The Linnaeansystem requires that the namesof some taxa indicate their rank. Speciesare given binomials, higher taxa have uninomials, and subspeciesare assignedtrinomials . So a species' binomial designation indicates its rank. But if the existence of the speciescategory is suspect, then so too is the practiceof using a taxon' s name to indicate that it is a species , for if there is no speciescategory , then no taxa should be designatedas " species." This is one potential problem with the Linnaeanrules of nomenclature. There are other problems with those rules, and they arise regardlessof whether one is a skeptic of the speciescategory. The bulk of this chapter outlines the problems facing the speciescategory and the rules of nomenclaturegoverning the naming of species.Those problems are seriousenough to consider abandoningthe entire Linnaeansystem. Replacing the Linnaeansystem with an alternative system is no small task, however, nor should it be done lightly . The Linnaean system is firmly entrenchedin biology , not to mention in popular culture. Biologists and philosophers must build a persuasivecasefor rejecting the Linnaeansystem, and they must develop viable alternatives. The last section of this chapter introduces some non-Linnaeansystemsworthy of consideration. In doing so, it offers a glimpse of what might be the future of biological nomenclature. THE D I STI NCrl ON BETWEENSPECIESAND mGHER TAXA The assumption that there is a fundamental ontological divide between lower taxa and higher taxa is an old one, dating back to Aristotle . This section examines several prominent criteria for distinguishing lower from ' higher taxa, starting. with Linnaeuss and ending with two contemporary suggestions. The criteria for distinguishing lower from higher taxa encounter a number of difficulties. This negative result castsa shadow on the distinction between speciesand higher taxa as well as on the existenceof the species categoryd . Boyd, chapter6 in this volume). Linnaean Species

For Linnaeus, the fundamental divide in his hierarchy lies between genera and all other higher taxa. Classificationsof speciesand genera reflect real groups in nature, whereas classificationsof classesand orders are artificial (Cain .1958, 148, 152- 153). This distinction stems from a central tenet of

286

/. Speaes Begone!

. According to Unnaeus, Linnaeus's biological theory: Aristotelian essentialism the membersof speciesand genera are endowed with essentialnatures created by God. The job of the biologist is to discover those essencesand their , on the associatedspeciesand genera (Ereshefsky1997). Orders and classes . artmcial constructs and are on other hand, are constructed pragmaticgrounds "An order is a subdivision of classesneededto avoid more placing together " Unnaeus in Mayr 1982, 175). follow the mind can than , ( quoted genera mind have and , Unnaeus to , independentessences genera species According us. to their existence owe and classes whereasorders ' When we dive a bit deeperinto Unnaeuss biological theory, in particular his sexual system, we see more clearly why he thought that speciesand . Following Cesalpino, , have essences genera, but not orders and classes Unnaeusbelieved that plants have two vital functions: nutrition , which preserves the individual, and reproduction, which preserves the kind (Larson " 1971, 146). The function of reproduction in plants is found in their fructifi " cation structures - namely, their flowers and fruits. Fructification structures are found at the level of genera, and are the essencesof genera, according to ' Linnaeus(Larson 1971, 74 ffj . Moving down a level, a species essenceis a ' combination of its genuss fructification structure and those parts involved in the function of nutrition (Larson 1971, 115 ff.). The essencesof speciesand they are responsible genera are particularly important for Linnaeusbecause 's God of taxa for the continued existence beyond original creation. Fructification structuresallow the membersof a speciesor a genusto reproduceand thus allow taxa to continue. Classesand orders, on the other hand, are not definedby fructification structures. They are merely aggregatesof organisms . containing different fructification structures ' s divide between In the last hundred years, Linnaeus genera and higher taxa has fallen on hard times. One reasonis the rejection of essentialismin Darwinian biology . Unnaeusthought that speciesand genera, but not orders and classes ; thus, the former are real, but not the latter. In , had essences . (See Darwinian biology , no taxon of any rank has a taxon-specmcessence Hull 1965 and Sober 1980 for arguments against taxa, especially species, ; d . Wilson, chapter 7 in this volume.) The relevancehere is having essences , so that speciesand generano more have essencesthan do orders and classes ' essentialist and lower between the distinction for Unnaeuss holding ground higher taxa no longer applies. Unnaeus also based his distinction on the assumption that only species and generahave generic-specificfructification structures. Orders and classes are artmcial groups of genera containing different fructification structures. Bio}ogists no longer believe that fructmcation structures are responsible for the existence of taxa. Instead, many hold that taxa are the result of interbreeding among conspecific organisms (see, for example, Mayr 1970, 3-73- 374, Eldredge and Cracraft 1980, 89- 90). Speciesare populations of organismsthat exchangegenetic material through interbreeding. That process causesthe local populations of a speciesto evolve as a unit. In contrast,

287

/ : Speciesand the UnnaeanHierarchy Ereshefsky

the speciesof a higher taxon do not exchangegenetic material, and the evolution of a higher taxon is merely the by -product of evolution occurring within its species. Thus, instead of Linnaeus's ontological divide between generaand other higher taxa, the authors of the Modem Synthesis, aswell as many contemporarybiologists, hold that the ontological divide lies between speciesand all higher taxa.1

The Units of Evolution In place of Linnaeus's distinction we have the speciesfhighertaxa distinction of the Modem Synthesis. Speciesare actively evolving entities, often referred to as the units of evolution. Higher taxa, including genera, are merely artifacts of processes occurring at lower levels of organization. This distinction is widely held by contemporary biologists (for example, Mayr 1970, 373374; Eldredge and Cracraft 1980, 89- 90, 249; Ghiselin 1987, 141). Typically , two sorts of arguments are given to support this distinction, each highlighting a processthat is supposedto be common and unique to speciestaxa. According to the first argument, gene flow is an essentialprocessfor the continued existenceof species. New traits that arise in one local population are spread by gene flow to the other local populations of a species. As a result, a speciesevolves as an evolutionary unit rather than an aggregateof local populations. No comparablemechanismto gene flow exists among the speciesof a higher taxon. The evolution of a higher taxon is merely a historical by -product of evolution occurring in its species. Species , therefore, are active agentsof evolution, while higher taxa are merely passiveresults of evolution. Though this basis for the speciesfhighertaxa distinction is quite popular, it has been called into question. A number of biologists argue that many specieslack the integrating force of gene flow. If they are correct, then many speciesare akin to higher taxa: they are merely aggregates of processes working at lower levels of biological organization (Ehrlich and Raven 1969, Grant 1980, Mishler and Donoghue 1982). Suppose, as many biologists do, that asexualorganismsform speciestaxa. The membersof such speciesare not bound by interbreeding, but by such processes as selection, genetic homeostasis , or developmentalcanalization(see Templeton 1989 for a discussion of these processes). Such processes causea group of organisms to belong to a single specieswithout requiring any causalinteraction among those organisms} Selection, for instance, can maintain the unity of a species by affecting individual membersin a similar fashion. The samegoes for the actions of genetic homeostasisand developmental canalization. If asexual , then such speciesare both structurally and causally organismsform species akin to higher taxa. One need not posit the existenceof asexualspeciesto find problems with the species / higher taxa distinction. Ehrlich and Raven (1969) and Templeton (1989) suggest that many speciesof sexual organismscontain local popula-

288

/ V. SpeciesBegone!

tions that exchangelittle or no geneticmaterial. Here, the claim is not that the local populations momentarily fail to exchange genetic material, but that they fail do so for a significant amount of time. If thesebiologists are right , then many speciesof sexual organismsare bound by processes working at lower levels of biological organization than at the level of the entire speciesd . Sterelny, chapter 5 in this volume). The unity of such speciesmay be the result of interbreedingwithin local populations, or their unity may be due to processes that independently affect organisms, such as selection or genetic . Either way, somespeciesof sexualorganismsare akin to higher homeostasis taxa: they are bound by causalmechanismsacting at lower levels of biological , organization. If such sexual speciesexist, or if there are asexualspecies from not then the processof gene flow does universally distinguish species higher taxa. Another argument for the speciesfhighertaxa distinction highlights the processof speciation. This argument can be found in Eldredge and Cracraft (1980, 327) and Mayr (1982, 296), and it runs as follows. Speciationis the primary causeof change in evolution. It occurs in speciesbut not higher taxa. Therefore, species,but not higher taxa, are the units of evolution. Once again, speciesare the active agents of evolution, whereas higher taxa are merely passive results. The trouble with this line of reasoning is that the locus of speciation is neither the species nor the higher taxon, but the founder population. According to Mayr' s (1970) allopatric model, speciation begins when a small population of organisms becomes isolated" and is . The population undergoes a genetic exposed to new selection pressures " and becomesthe revolution founding population of a new species. The point here is that the processof speciation occurs in founding populations , not in entire species. So if higher taxa are not evolutionary units becausespeciation occurs in only a portion of them (founder populations), then by parallel reasoning, speciesare not evolutionary units either. Now one might counter that it is inaccurate to assert that speciation occurs in founder populations rather than entire speciesbecausethere is nothing more to an incipient speciesthan its founder population. Suppose we grant that point . Then it applies to incipient higher taxa as well: at some stage in their developmentthey are nothing more than founder populations. Thus, if speciesare evolutionary units becausespeciation occurs in their incipient forms, then higher taxa are evolutionary units as well. In brief, the processof speciationdoes not separatespeciesfrom higher taxa. Stepping back, we seethat the distinction between speciesand higher taxa is problematic. Linnaeus drew the distinction along the lines of essential and fructification systems. Modem biology has rejected essentialism natures ' ' and Linnaeuss sexualsystem. Alternatively , the authors of the Modem Synthesis and many contemporaryauthors basetheir distinction betweenspecies and higher taxa on the processes of speciationand interbreeding, but those processes do not adequately distinguish speciesfrom higher taxa, either.

1.89

/ : Species andthe Unnaean Ereshefsky Hierarchy

Perhapssomeyet to be discoveredprocessor feature will clearly distinguish speciesfrom higher taxa. That certainly is a possibility. But until we know of sucha process, we should maintain a healthy skepticismtoward the species / higher taxa distinction. Doubt concerning the species / higher taxa distinction affects our confidence in the reality of the speciescategory. A minimal requirement for believing in the existenceof a category is having sufficient grounds for distinguishin it from its neighboring categories. If no biological criteriaade quately distinguish speciestaxa from higher taxa, then we lack grounds for believing that species, genera, and orders are ontologically distinct categories . It is worth emphasizingthat the argument here is merely against the existenceof the Linnaeancategories, especially the speciescategory. Nothing I have said castsdoubt on the reality of particular taxa. We can remain confident in the existenceof such taxa as Homo and Homo sapienseven if the Linnaeancategoriesgo by the waysided . Mishler, chapter 12 in this volume).

SPECIES PLURALISM Reasonsfor doubting the existenceof the speciescategory come from other quarters as well. The current taxonomic literature contains no less than a dozen speciesconcepts(seeEreshefsky1992b and recent issuesof Systematic Biology). Biologists and philosophershave respondedto this wealth of concepts in two ways. Monists believe that biologists should settle on a single correct concept. Pluralists suggest that a number of speciesconceptsshould be accepted as legitimate (cE. Dupre and Hull, chapters 1 and 2 in this volume).3 Undoubtedly, a number of currently proposed speciesconcepts will be found wanting and relegatedto the history of science. However, two major - the approaches to species interbreeding and phylogenetic- are currently well entrenchedin biology for good theoretical and empirical reasons(de Queiroz and Donoghue 1988, Ereshefsky 1992a). The interbreeding and phylogenetic approaches highlight noncomparabletypes of speciestaxa, so if one acceptsthese two approaches, then there is no single unitary species category, but a heterogeneouscollection of basetaxa referred to by the term . Speciespluralism, in other words, poses a threat to the existenceof species the speciescategory. The aim of this section is to display that threat and to highlight the disunity of the speciescategory. But before getting to that, we need a quick introduction to the interbreedingand phylogenetic approaches. The interbreedingapproachis typified by Mayr ' s (1970) biological species concept. Speciesare gene pools held together by interbreeding and protected . Examples of reproductive by various reproductive isolating mechanisms isolating mechanismsinclude courtship behavior that prevents the mating of two interspecific organismsand hybrid inviability if such mating does occur. Other speciesconceptsthat fall under the general interbreeding

290

v . Sp/ aesBegone !

' approach include Ghiselin s (1974) reproductive competition concept and ' Patersons (1985) mate recognition concept. Though theseconceptsdiffer in " important respects, they agree that a speciesis a field for genetic recombination " Carson 1957). ( The phylogeneticapproachis found in the work of cladists (for example, Cracraft 1983, Mishler and Brandon 1987, Ridley 1989). Propinquity of descentis the operative notion here. A taxon must contain a single ancestral speciesas well as all and only its descendantspecies. Any taxon meeting . The founder of cladism, Willi Hennig, did that requirement is monophyletic not intend the notion of monophyly to apply to species , but reservedit for higher taxa. Recent cladists have extended its use, however. Mishler and " Brandon(1987, 46), for example, define a speciesas the least inclusive taxon recognized in a classificationinto which organismsare grouped becauseof " degreeof monophyly (seealso Mishler, chapter 12 in this volume). Phylogenetic species are base monophyletic taxa maintained by a number of forces- including selection, interbreeding, genetic homeostasis , and developmental canalization. Given thesetwo approaches to species- the interbreeding and the phylo genetic- one might wonder what feature unifies phylogenetic and interbreeding speciesinto a single categoryd . de Queiroz, chapter 3 in this volume). A simple requirement for the existence of a category is that its entities sharea feature that distinguishes them from entities in other categories . If phylogenetic and interbreeding speciesare both species, then they should share some common and distinctive feature. If they lack such a feature , then the species category consists of noncomparable entities.4 The remainder of this section examinesthose features that might render interbreeding and phylogenetic speciescomparable. A good place to start is with the processes that maintain the existenceof interbreedingand phylogenetic species.In interbreeding species,gene flow is the primary unifying force. How much and how often genetic material must be exchanged varies from interbreeding species to interbreeding species , the situation is quite ( Templeton 1989, 165). For many phylogenetic species different. Some phylogenetic species contain sexual organisms living in isolated populations. In such species, selection, genetic homeostasis , or flow. not forces are the canalization , gene primary unifying developmental Similarly, phylogenetic speciesconsisting of asexualorganismsare bound by forces other than interbreeding, so different types of speciesare unified by different types of processes. Consequently, no single unifying process(or set of processes) servesas the common feature of all speciestaxa.S P~rhaps we would be better off looking at the structuresof interbreeding and phylogenetic species.Perhapsa significant similarity can be found there. So~ e interbreeding speciesconsist of a single local population. Other interbreeding speciesconsist of a number of local populations connectedby inter to have such causes of the species bree4ing. Either way, process interbreeding the structure of causallyintegrated entities. Of course, interbreeding among

291

.

/ : Species andthe Linnaean Hierarchy Ereshefsky

the membersof a speciesis not a continuous affair, but contains temporal gaps. Moreover, many membersof a speciesmay not breed at all (though they are the result of interbreeding). Nonetheless, the membersof such species are bound into single speciesby the interactive force of interbreeding. The structure of many phylogenetic speciesis different. Some are base monophyletic taxa of asexualorganisms. The forces that maintain the existence of such speciesare selection, genetic homeostasis , and developmental canalization. These forces can maintain a group of organisms as a species without requiring causalinteraction among those organisms. Selection, for example, can maintain the unity of a speciesby individually affecting its organisms. Similarly, homeostatic genotypes separatelycauseorganismsto have certain traits, and developmentalcanalizationindependently constrains the ontogeny of each organism. Such phylogenetic speciesare not causally integrated entities. They are the result of independent forces acting at the level of the organism. The sameobservation can be made of phylogenetic speciesconsisting of sexualorganismsliving in isolated populations. In such , interbreeding may preserve unity at the level of local populations, species but the preservation of entire speciesis due to selection, genetic homeostasis, or developmentalcanalization(Ehrlich and Raven 1969, El dredge and Gould 1972, Mishler and Donoghue 1982). Stepping back, we see that interbreeding speciesare causally integrated entities, whereasmany phylogenetic speciesare not. Interbreeding species are bound by interactive processes that occur at the level of entire ; species many phylogenetic speciesare the result of noninteractive processes working at lower levels of biological organization. These considerationslead to the conclusion that interbreeding speciesand many phylogenetic species have different ontological structures. In other words, the speciescategory is an ontologically mixed bag of entities. So far we have seen that speciestaxa are bound by different types of processes and that speciestaxa have very different ontological structures. The heterogeneousnature of the speciescategory is brought into sharper focus when we seethat lineagesthat are consideredspeciesin one approach often fail to be speciesin the other. For example, some interbreeding species fail to form phylogenetic species because they do not contain all the descendantsof a common ancestor. The &eshwater fish group Xiphophorus contains a seriesof populations (figure 11.1). The membersof C and F successfullyinterbreed and are reproductively isolated &om the membersof the other populations. Thus C + F forms a single interbreeding species(Rosen 1979, 275- 279). Yet on the phylogenetic approach, C + F cannot be a single species becauseit does not contain all the descendantsof the common ancestorX. In the phylogenetic approach, C and F are two distinct species. This type of discrepancyis far &om unusual. Consider ancestralspecies. Supposean interbreeding speciesA spawnsa new speciesC; yet A continues to exist as B (figure 11.2). In the interbreeding approach, A + B forms one , butCforms another. In the phylogenetic approach, A + B cannot be species

292

v . SpeciesBegone!

IF IDE ABC X (afterRosen1979, Figure 11.1 A simplifiedcladogramof the swordBshgroupXiphophorus 276). C + F forms a spedeson the interbreedingapproachbut not on the phylogenetic . approach

, whereas C forms , A + B forms one species Figure 11. 2 On the interbreeding approach another species.On the phylogenetic approach , A, B, and C are three different species.

a species because it doe ~ not contain all the descendants of the common ancestor X : A + B is missing the organisms of C. So, for many cladists, A goes extinct at the time of speciation and two new species, B and C , take its place. Phylogenetic species often fail to be interbreeding species as well . Some proponents of the phylogenetic approach believe that asexual organisms fon. n species (Mishler and Brandon 1987 ). Supporters of the interbreeding approach disagree and argue that asexual organisms do not belong to any sP.ecies (Ghiselin 1987 , Hull 1987 ). For my part , I see no reason why the term speciesshould be reserved for only sexual organisms (see Ereshefsky 1992b and 1998 for arguments ). If asexual organisms do form phylogenetic species, then many phylogenetic species fail to be interbreeding species. This

293

~ . Hierarchy Ereshefsky:Speaesandthe Linnaean

sort of problem does not depend on the assumpHon that asexualorganisms form species.It also occursfor somegroups of sexualorganisms. As we have seen, some phylogeneH